1<!DOCTYPE html> 2<html lang="en"><head>
3<script src="index_files/libs/clipboard/clipboard.min.js"></script>
vendor: 1 bytes, line 3
3
4<script src="index_files/libs/quarto-html/tabby.min.js"></script>
vendor: 1 bytes, line 4
4
5<script src="index_files/libs/quarto-html/popper.min.js"></script>
vendor: 1 bytes, line 5
5
6<script src="index_files/libs/quarto-html/tippy.umd.min.js"></script>
6 7<link href="index_files/libs/quarto-html/tippy.css" rel="stylesheet"> 8<link href="index_files/libs/quarto-html/light-border.css" rel="stylesheet"> 9<link href="index_files/libs/quarto-html/quarto-syntax-highlighting-ea385d0e468b0dd5ea5bf0780b1290d9.css" rel="stylesheet" id="quarto-text-highlighting-styles">
10<script src="index_files/libs/quarto-contrib/jquery-3.6.3/jquery-3.6.3.min.js"></script>
vendor: 1 bytes, line 10
10
11<script src="index_files/libs/quarto-contrib/code-fullscreen-1.0.0/code-fullscreen.js" defer="true"></script>
11 12<link href="index_files/libs/quarto-contrib/code-fullscreen-1.0.0/code-fullscreen.css" rel="stylesheet"> 13<link href="index_files/libs/quarto-contrib/fontawesome6-6.7.2/all.min.css" rel="stylesheet"> 14<link href="index_files/libs/quarto-contrib/fontawesome6-6.7.2/latex-fontsize.css" rel="stylesheet">
15<script src="index_files/libs/quarto-contrib/glightbox/glightbox.min.js"></script>
15 16<link href="index_files/libs/quarto-contrib/glightbox/glightbox.min.css" rel="stylesheet"> 17<link href="index_files/libs/quarto-contrib/glightbox/lightbox.css" rel="stylesheet"><meta charset="utf-8"> 18 <meta name="generator" content="quarto-1.7.33"> 19 20 <meta name="author" content="Sakeef M. Karim Amherst College AN INTRODUCTION TO QUANTITATIVE SOCIOLOGYâCULTURE & POWER"> 21 <title>Lecture 12 for SOCIâ269 at Amherst College</title> 22 <meta name="apple-mobile-web-app-capable" content="yes"> 23 <meta name="apple-mobile-web-app-status-bar-style" content="black-translucent"> 24 <meta name="viewport" content="width=device-width, initial-scale=1.0, maximum-scale=1.0, user-scalable=no, minimal-ui"> 25 <link rel="stylesheet" href="index_files/libs/revealjs/dist/reset.css"> 26 <link rel="stylesheet" href="index_files/libs/revealjs/dist/reveal.css"> 27 <style> 28 code{white-space: pre-wrap;} 29 span.smallcaps{font-variant: small-caps;} 30 div.columns{display: flex; gap: min(4vw, 1.5em);} 31 div.column{flex: auto; overflow-x: auto;} 32 div.hanging-indent{margin-left: 1.5em; text-indent: -1.5em;} 33 ul.task-list{list-style: none;} 34 ul.task-list li input[type="checkbox"] { 35 width: 0.8em; 36 margin: 0 0.8em 0.2em -1em; /* quarto-specific, see https://github.com/quarto-dev/quarto-cli/issues/4556 */ 37 vertical-align: middle; 38 } 39 /* CSS for citations */ 40 div.csl-bib-body { } 41 div.csl-entry { 42 clear: both; 43 } 44 .hanging-indent div.csl-entry { 45 margin-left:2em; 46 text-indent:-2em; 47 } 48 div.csl-left-margin { 49 min-width:2em; 50 float:left; 51 } 52 div.csl-right-inline { 53 margin-left:2em; 54 padding-left:1em; 55 } 56 div.csl-indent { 57 margin-left: 2em; 58 } </style> 59 <link rel="stylesheet" href="index_files/libs/revealjs/dist/theme/quarto-af1b25da374d2d51a4bce390b12b8811.css"> 60 <link rel="stylesheet" href="logo.css"> 61
61<script>window.backupDefine = window.define; window.define = undefined;</script>
61<script src="https://cdn.jsdelivr.net/npm/katex@latest/dist/katex.min.js"></script>
61 62
62<script>document.addEventListener("DOMContentLoaded", function () { 63 var mathElements = document.getElementsByClassName("math"); 64 var macros = []; 65 for (var i = 0; i < mathElements.length; i++) { 66 var texText = mathElements[i].firstChild; 67 if (mathElements[i].tagName == "SPAN") { 68 katex.render(texText.data, mathElements[i], { 69 displayMode: mathElements[i].classList.contains('display'), 70 throwOnError: false, 71 macros: macros, 72 fleqn: false 73 }); 74}}}); 75 </script>
75 76
76<script>window.define = window.backupDefine; window.backupDefine = undefined;</script>
76<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@latest/dist/katex.min.css"> 77 <link href="index_files/libs/revealjs/plugin/quarto-line-highlight/line-highlight.css" rel="stylesheet"> 78 <link href="index_files/libs/revealjs/plugin/reveal-menu/menu.css" rel="stylesheet"> 79 <link href="index_files/libs/revealjs/plugin/reveal-menu/quarto-menu.css" rel="stylesheet"> 80 <link href="index_files/libs/revealjs/plugin/reveal-chalkboard/font-awesome/css/all.css" rel="stylesheet"> 81 <link href="index_files/libs/revealjs/plugin/reveal-chalkboard/style.css" rel="stylesheet"> 82 <link href="index_files/libs/revealjs/plugin/quarto-support/footer.css" rel="stylesheet"> 83 <style type="text/css"> 84 .reveal div.sourceCode { 85 margin: 0; 86 overflow: auto; 87 } 88 .reveal div.hanging-indent { 89 margin-left: 1em; 90 text-indent: -1em; 91 } 92 .reveal .slide:not(.center) { 93 height: 100%; 94 overflow-y: auto; 95 } 96 .reveal .slide.scrollable { 97 overflow-y: auto; 98 } 99 .reveal .footnotes { 100 height: 100%; 101 overflow-y: auto; 102 } 103 .reveal .slide .absolute { 104 position: absolute; 105 display: block; 106 } 107 .reveal .footnotes ol { 108 counter-reset: ol; 109 list-style-type: none; 110 margin-left: 0; 111 } 112 .reveal .footnotes ol li:before { 113 counter-increment: ol; 114 content: counter(ol) ". "; 115 } 116 .reveal .footnotes ol li > p:first-child { 117 display: inline-block; 118 } 119 .reveal .slide ul, 120 .reveal .slide ol { 121 margin-bottom: 0.5em; 122 } 123 .reveal .slide ul li, 124 .reveal .slide ol li { 125 margin-top: 0.4em; 126 margin-bottom: 0.2em; 127 } 128 .reveal .slide ul[role="tablist"] li { 129 margin-bottom: 0; 130 } 131 .reveal .slide ul li > *:first-child, 132 .reveal .slide ol li > *:first-child { 133 margin-block-start: 0; 134 } 135 .reveal .slide ul li > *:last-child, 136 .reveal .slide ol li > *:last-child { 137 margin-block-end: 0; 138 } 139 .reveal .slide .columns:nth-child(3) { 140 margin-block-start: 0.8em; 141 } 142 .reveal blockquote { 143 box-shadow: none; 144 } 145 .reveal .tippy-content>* { 146 margin-top: 0.2em; 147 margin-bottom: 0.7em; 148 } 149 .reveal .tippy-content>*:last-child { 150 margin-bottom: 0.2em; 151 } 152 .reveal .slide > img.stretch.quarto-figure-center, 153 .reveal .slide > img.r-stretch.quarto-figure-center { 154 display: block; 155 margin-left: auto; 156 margin-right: auto; 157 } 158 .reveal .slide > img.stretch.quarto-figure-left, 159 .reveal .slide > img.r-stretch.quarto-figure-left { 160 display: block; 161 margin-left: 0; 162 margin-right: auto; 163 } 164 .reveal .slide > img.stretch.quarto-figure-right, 165 .reveal .slide > img.r-stretch.quarto-figure-right { 166 display: block; 167 margin-left: auto; 168 margin-right: 0; 169 } 170 </style> 171 <link rel="icon" type="media/png" href="media/line-chart.png"> 172 <link href="https://cdn.jsdelivr.net/npm/[email protected]/dist/css/bootstrap.min.css" rel="stylesheet" integrity="sha384-EVSTQN3/azprG1Anm3QDgpJLIm9Nao0Yz1ztcQTwFspd3yD65VohhpuuCOmLASjC" crossorigin="anonymous"> 173</head> 174<body class="quarto-light"> 175 <div class="reveal"> 176 <div class="slides"> 177 178<section id="title-slide" class="quarto-title-block center"> 179 <h1 class="title"><p><span style="color:#a9a9a9;">Week 12</span><br> 180Machine Learning in<br> 181Python <svg aria-hidden="true" role="img" viewbox="0 0 448 512" style="height:1em;width:0.88em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:#5d3c85;overflow:visible;position:relative;"><path d="M439.8 200.5c-7.7-30.9-22.3-54.2-53.4-54.2h-40.1v47.4c0 36.8-31.2 67.8-66.8 67.8H172.7c-29.2 0-53.4 25-53.4 54.3v101.8c0 29 25.2 46 53.4 54.3 33.8 9.9 66.3 11.7 106.8 0 26.9-7.8 53.4-23.5 53.4-54.3v-40.7H226.2v-13.6h160.2c31.1 0 42.6-21.7 53.4-54.2 11.2-33.5 10.7-65.7 0-108.6zM286.2 404c11.1 0 20.1 9.1 20.1 20.3 0 11.3-9 20.4-20.1 20.4-11 0-20.1-9.2-20.1-20.4.1-11.3 9.1-20.3 20.1-20.3zM167.8 248.1h106.8c29.7 0 53.4-24.5 53.4-54.3V91.9c0-29-24.4-50.7-53.4-55.6-35.8-5.9-74.7-5.6-106.8.1-45.2 8-53.4 24.7-53.4 55.6v40.7h106.9v13.6h-147c-31.1 0-58.3 18.7-66.8 54.2-9.8 40.7-10.2 66.1 0 108.6 7.6 31.6 25.7 54.2 56.8 54.2H101v-48.8c0-35.3 30.5-66.4 66.8-66.4zm-6.7-142.6c-11.1 0-20.1-9.1-20.1-20.3.1-11.3 9-20.4 20.1-20.4 11 0 20.1 9.2 20.1 20.4s-9 20.3-20.1 20.3z"></path></svg></p></h1> 182 <p class="subtitle"><span style="color:#9c9c9c; font-family: Dynalight; font-size: 130%;">Sociâ269</span></p> 183 184<div class="quarto-title-authors"> 185<div class="quarto-title-author"> 186<div class="quarto-title-author-name"> 187Sakeef M. Karim <br><span style="font-size: 80%; color:#9c9c9c;"><em>Amherst College</em></span><br><a href="https://www.sakeefkarim.com" class="external" target="_blank"><svg aria-hidden="true" role="img" viewbox="0 0 512 512" style="height:1em;width:1em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:#5d3c85;overflow:visible;position:relative;"><path d="M352 256c0 22.2-1.2 43.6-3.3 64H163.3c-2.2-20.4-3.3-41.8-3.3-64s1.2-43.6 3.3-64H348.7c2.2 20.4 3.3 41.8 3.3 64zm28.8-64H503.9c5.3 20.5 8.1 41.9 8.1 64s-2.8 43.5-8.1 64H380.8c2.1-20.6 3.2-42 3.2-64s-1.1-43.4-3.2-64zm112.6-32H376.7c-10-63.9-29.8-117.4-55.3-151.6c78.3 20.7 142 77.5 171.9 151.6zm-149.1 0H167.7c6.1-36.4 15.5-68.6 27-94.7c10.5-23.6 22.2-40.7 33.5-51.5C239.4 3.2 248.7 0 256 0s16.6 3.2 27.8 13.8c11.3 10.8 23 27.9 33.5 51.5c11.6 26 20.9 58.2 27 94.7zm-209 0H18.6C48.6 85.9 112.2 29.1 190.6 8.4C165.1 42.6 145.3 96.1 135.3 160zM8.1 192H131.2c-2.1 20.6-3.2 42-3.2 64s1.1 43.4 3.2 64H8.1C2.8 299.5 0 278.1 0 256s2.8-43.5 8.1-64zM194.7 446.6c-11.6-26-20.9-58.2-27-94.6H344.3c-6.1 36.4-15.5 68.6-27 94.6c-10.5 23.6-22.2 40.7-33.5 51.5C272.6 508.8 263.3 512 256 512s-16.6-3.2-27.8-13.8c-11.3-10.8-23-27.9-33.5-51.5zM135.3 352c10 63.9 29.8 117.4 55.3 151.6C112.2 482.9 48.6 426.1 18.6 352H135.3zm358.1 0c-30 74.1-93.6 130.9-171.9 151.6c25.5-34.2 45.2-87.7 55.3-151.6H493.4z"></path></svg></a> <a href="mailto:[email protected]"><svg aria-hidden="true" role="img" viewbox="0 0 512 512" style="height:1em;width:1em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:#5d3c85;overflow:visible;position:relative;"><path d="M48 64C21.5 64 0 85.5 0 112c0 15.1 7.1 29.3 19.2 38.4L236.8 313.6c11.4 8.5 27 8.5 38.4 0L492.8 150.4c12.1-9.1 19.2-23.3 19.2-38.4c0-26.5-21.5-48-48-48H48zM0 176V384c0 35.3 28.7 64 64 64H448c35.3 0 64-28.7 64-64V176L294.4 339.2c-22.8 17.1-54 17.1-76.8 0L0 176z"></path></svg></a> <br> <span style="font-weight: 300; font-size: 60%; color: #9c9c9c;">AN INTRODUCTION TO QUANTITATIVE SOCIOLOGYâCULTURE & POWER</span> 188</div> 189</div> 190</div> 191 192</section> 193<section> 194<section id="a-very-gentle-introduction-november-17th" class="title-slide slide level1 center" data-background-color="#43464B"> 195<h1>A <span style="text-decoration: underline;"><em>Very</em></span> Gentle Introductionâ<br>November 17<span style="font-size: 70%;"><sup>th</sup></span></h1> 196 197</section> 198<section id="reminders" class="slide level2"> 199<h2>Reminders</h2> 200<div class="centre-text"> 201<p>You have a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>bit</em></strong></span> more time with<br>the second coding assignment.</p> 202</div> 203</section> 204<section id="reminders-1" class="slide level2"> 205<h2>Reminders</h2> 206<div style="margin-top: 90px; text-align: center;"> 207<div class="callout callout-warning callout-titled callout-style-default"> 208<div class="callout-body"> 209<div class="callout-title"> 210<div class="callout-icon-container"> 211<i class="callout-icon"></i> 212</div> 213<p><strong>Second Coding Assignment <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>Soft</em></strong></span> Deadline</strong></p> 214</div> 215<div class="callout-content"> 216<p>Your second coding assignment is due by <strong>8:00 PM</strong>
216 on Friday, November 21<span style="font-size:70%"><sup>st</sup></span>.<br>This is a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>soft</em></strong></span> deadline.</p> 217</div> 218</div> 219</div> 220</div> 221</section> 222<section id="reminders-2" class="slide level2"> 223<h2>Reminders</h2> 224<div style="margin-top: 90px; text-align: center;"> 225<div class="callout callout-important callout-titled callout-style-default"> 226<div class="callout-body"> 227<div class="callout-title"> 228<div class="callout-icon-container"> 229<i class="callout-icon"></i> 230</div> 231<p><strong>Second Coding Assignment <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>Hard</em></strong></span> Deadline</strong></p> 232</div> 233<div class="callout-content"> 234<p>All coding assignments (in <code>Python</code> <svg aria-hidden="true" role="img" viewbox="0 0 448 512" style="height:1em;width:0.88em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:#5d3c85;overflow:visible;position:relative;"><path d="M439.8 200.5c-7.7-30.9-22.3-54.2-53.4-54.2h-40.1v47.4c0 36.8-31.2 67.8-66.8 67.8H172.7c-29.2 0-53.4 25-53.4 54.3v101.8c0 29 25.2 46 53.4 54.3 33.8 9.9 66.3 11.7 106.8 0 26.9-7.8 53.4-23.5 53.4-54.3v-40.7H226.2v-13.6h160.2c31.1 0 42.6-21.7 53.4-54.2 11.2-33.5 10.7-65.7 0-108.6zM286.2 404c11.1 0 20.1 9.1 20.1 20.3 0 11.3-9 20.4-20.1 20.4-11 0-20.1-9.2-20.1-20.4.1-11.3 9.1-20.3 20.1-20.3zM167.8 248.1h106.8c29.7 0 53.4-24.5 53.4-54.3V91.9c0-29-24.4-50.7-53.4-55.6-35.8-5.9-74.7-5.6-106.8.1-45.2 8-53.4 24.7-53.4 55.6v40.7h106.9v13.6h-147c-31.1 0-58.3 18.7-66.8 54.2-9.8 40.7-10.2 66.1 0 108.6 7.6 31.6 25.7 54.2 56.8 54.2H101v-48.8c0-35.3 30.5-66.4 66.8-66.4zm-6.7-142.6c-11.1 0-20.1-9.1-20.1-20.3.1-11.3 9-20.4 20.1-20.4 11 0 20.1 9.2 20.1 20.4s-9 20.3-20.1 20.3z"></path></svg>) <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>must</em></strong></span> be submitted by <strong>8:00 PM</strong> on Wednesday, December 3<span style="font-size:70%"><sup>rd</sup></span>.</p> 235</div> 236</div> 237</div> 238</div> 239</section> 240<section id="reminders-3" class="slide level2"> 241<h2>Reminders</h2> 242<div style="text-align: center; margin-top: 25px;"> 243<p>Once again, assignment instructions are available online.</p> 244<p align="center"> 245<iframe class="iframe" src="https://soci-269-a2.netlify.app/" height="450" width="1040" style="border: 1px solid #2e3846;"> 246</iframe> 247</p> 248<div style="text-align: center; font-size: 75%; margin-top:-40px"> 249<p><a href="https://soci-269-a1.netlify.app/" data-preview-link="true">Full Page</a></p> 250</div> 251</div> 252</section> 253<section id="an-update" class="slide level2"> 254<h2>An Update</h2> 255<h4 id="final-presentation-and-term-paper"><span style="color:#a9a9a9;">Final Presentation and Term Paper</span></h4> 256<div style="text-align: center; font-size: 87%; margin-top: 20px; margin-bottom: 20px;"> 257<p>Guidelines for your <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>final presentation</em></strong></span> and <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>term paper</em></strong></span> are live.</p> 258<p align="center"> 259<iframe class="iframe" src="https://soci-269-final.netlify.app/" height="450" width="1040" style="border: 1px solid #2e3846;"> 260</iframe> 261</p> 262<div style="text-align: center; font-size: 60%; margin-top:-40px"> 263<p><a href="https://soci-269-final.netlify.app/" data-preview-link="true">Full Page</a></p> 264</div> 265</div> 266</section></section> 267<section> 268<section id="what-is-machine-learning" class="title-slide slide level1 center" data-background-color="#43464B"> 269<h1>What is Machine Learning?</h1> 270 271</section> 272<section id="some-concrete-definitions" class="slide level2"> 273<h2>Some Concrete Definitions</h2> 274<div style="margin-top: 50px; font-size: 95%;"> 275<div class="fragment"> 276<p>Grimmer and colleagues <span class="citation" data-cites="grimmer2021">(<a href="#/references" role="doc-biblioref" onclick="">2021:396</a>
276, <span style="font-size: 55%; font-family: Red Hat Display; color: white; background-color: #5d3c85"><strong>EMPHASIS ADDED</strong></span>)</span></p> 277<div class="shrinker-plus"> 278<blockquote> 279<p>A class of flexible algorithmic and statistical techniques for <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>prediction</strong></span> and <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>dimension reduction</strong></span>.</p> 280</blockquote> 281</div> 282</div> 283<div style="margin-top: 40px;"> 284 285</div> 286<div class="fragment"> 287<p>Molina and Garip <span class="citation" data-cites="molina_machine_2019">(<a href="#/references" role="doc-biblioref" onclick="">2019:28</a>, <span style="font-size: 55%; font-family: Red Hat Display; color: white; background-color: #5d3c85"><strong>EMPHASIS ADDED</strong></span>)</span></p> 288<div class="shrinker-plus"> 289<blockquote> 290<p>A way to learn from data and estimate complex functions that <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>discover representations of some input</strong> (<span class="math inline">X</span>)</span>, or <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>link the input to an output</strong> (<span class="math inline">Y</span>)</span> in order to <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>make predictions on new data</strong></span>.</p> 291</blockquote> 292</div> 293</div> 294</div> 295</section> 296<section id="supervised-machine-learning-sml" class="slide level2 smaller"> 297<h2>Supervised Machine Learning <span style="font-size: 70%;">(SML)</span></h2> 298<h3 id="high-level-overview"><span style="color:#9c9c9c; font-size: 85%;">High-Level Overview</span></h3> 299<div style="margin-top: 40px;"> 300 301</div> 302<ul> 303<li class="fragment"><p>Once deployed, SML algorithms learn the complex patterns linking <span class="math inline">X</span>âa set of <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>features</em></span> (or independent variables)âto a <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>target variable</em></span> (or outcome), <span class="math inline">Y</span>.</p> 304<ul> 305<li class="fragment"><p><span style="font-size: 85%;">The goal of SML is to <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>optimize predictions</strong></span>âi.e., to find functions or algorithms that offer substantial predictive power when confronted with <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>new or unseen data</em></span>.</span></p></li> 306<li class="fragment"><p><span style="font-size: 85%;">Examples of SML algorithms include <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>logistic regressions</strong></span>, <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>random forests</strong></span>, <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>ridge regressions</strong></span>, <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>support vector machines</strong></span> and <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>neural networks</strong></span>.</span></p></li> 307</ul></li> 308<li class="fragment"><p>A quick note on terminology</p> 309<ul> 310<li class="fragment"><p><span style="font-size: 85%;">If a target variable is <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>quantitative</em></span>, we are dealing with a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>regression</strong></span> problem.</span></p></li> 311<li class="fragment"><p><span style="font-size: 85%;">If a target variable is <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>qualitative</em></span>, we are dealing with a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>classification</strong></span> problem.</span></p></li> 312</ul></li> 313</ul> 314</section> 315<section id="supervised-machine-learning-sml-cont." class="slide level2 smaller"> 316<h2>Supervised Machine Learning <span style="font-size: 70%;">(SML)</span> <em>cont</em>.</h2> 317<h3 id="an-illustration"><span style="color:#9c9c9c; font-size: 85%;">An Illustration</span></h3> 318<div class="panel-tabset"> 319<ul id="tabset-1" class="panel-tabset-tabby"><li><a data-tabby-default="" href="#tabset-1-1">(1) Feed Input Data</a></li><li><a href="#tabset-1-2">(2) Develop Algorithm</a></li><li><a href="#tabset-1-3">(3) Predictions on New Data</a></li></ul> 320<div class="tab-content"> 321<div id="tabset-1-1"> 322<div style="margin-top: 20px;"> 323 324</div> 325<div class="cell styled-output" data-layout-align="center"> 326<div class="cell-output cell-output-stdout"> 327<pre><code># A tibble: 80,000 Ã 6 328 fruit shape weight_g colour texture origin 329 <chr> <chr> <dbl> <chr> <chr> <chr> 330 1 apple spherical 150 red crispy china 331 2 banana curved 120 yellow creamy ecuador 332 3 orange spherical 148 orange juicy egypt 333 4 watermelon spherical 4500 green juicy spain 334 5 strawberry conical 13 red juicy mexico 335 6 grape spherical 5 green juicy chile 336 7 mango ellipsoidal 240 yellow juicy india 337 8 pineapple conical 2100 yellow juicy costa rica 338 9 apple spherical 140 green crispy usa 33910 banana curved 110 yellow creamy ecuador 340# â¹ 79,990 more rows</code></pre> 341</div> 342</div> 343</div> 344<div id="tabset-1-2"> 345<div style="text-align: center; margin-top: 10px;"> 346<div class="quarto-figure quarto-figure-center"> 347<figure> 348<p><a href="./media/neural_network.png" class="lightbox" data-gallery="quarto-lightbox-gallery-1" title="via scikit-learn"><img data-src="./media/neural_network.png" style="width:32.0%" alt="via scikit-learn"></a></p> 349<figcaption><a href="https://scikit-learn.org/stable/modules/neural_networks_supervised.html" class="external" target="_blank">via <span style="font-size: 85%;"><code>scikit-learn</code></span></a></figcaption> 350</figure> 351</div> 352</div> 353</div> 354<div id="tabset-1-3"> 355<div class="columns"> 356<div class="column" style="width:55%;"> 357<div class="fragment"> 358<div class="cell styled-output"> 359<div class="cell-output cell-output-stdout"> 360<pre><code># A tibble: 1 Ã 6 361 fruit shape weight_g colour texture origin 362 <chr> <chr> <int> <chr> <chr> <chr> 3631 <NA> conical 6 red juicy china </code></pre> 364</div> 365</div> 366</div> 367<div class="fragment"> 368<div class="cell styled-output"> 369<div class="cell-output cell-output-stdout"> 370<pre><code># A tibble: 8 Ã 2 371 fruit probability 372 <chr> <dbl> 3731 apple 0.03 3742 banana 0 3753 orange 0.05 3764 watermelon 0 3775 strawberry 0.71 3786 grape 0.21 3797 mango 0 3808 pineapple 0 </code></pre> 381</div> 382</div> 383</div> 384</div><div class="column" style="width:45%;"> 385<div class="fragment"> 386<div class="cell"> 387<div class="cell-output cell-output-stdout"> 388<pre><code># A tibble: 1 Ã 1 389 prediction 390 <chr> 3911 strawberry</code></pre> 392</div> 393</div> 394<p><img data-src="./media/strawberry.png" style="width:40.0%"></p> 395<div class="attribution"> 396<p><a href="https://www.flaticon.com/free-icons/strawberry" title="strawberry icons" style="font-size: 50%;">Strawberry icons created by BZZRINCANTATION - Flaticon</a></p> 397</div> 398</div> 399</div></div> 400</div> 401</div> 402</div> 403</section> 404<section id="unsupervised-machine-learning-uml" class="slide level2 smaller"> 405<h2>Unsupervised Machine Learning <span style="font-size: 70%;">(UML)</span></h2> 406<h3 id="high-level-overview-1"><span style="color:#9c9c9c; font-size: 85%;">High-Level Overview</span></h3> 407<ul> 408<li class="fragment"><p>UML techniques search for a representation of the inputs (or features) that is more useful than <span class="math inline">X</span> itself <span class="citation" data-cites="molina_machine_2019">(<a href="#/references" role="doc-biblioref" onclick="">
408Molina and Garip 2019</a>)</span>.</p> 409<ul> 410<li class="fragment"><span style="font-size: 95%;">Put another way, UML algorithms search for <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>hidden structure or latent patterns in high dimensional space</strong></span>.</span></li> 411</ul></li> 412<li class="fragment"><p>In UML, there is no <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>observed</em></span> outcome variable <span class="math inline">Y</span>âor targetâto <em>supervise</em> the estimation process. Instead, we only have a vector of inputs to work with.</p></li> 413<li class="fragment"><p>The goal in UML is to develop a lower-dimensional representation of complex data by inductively learning from the <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>interrelationships among inputs</em></strong></span>.</p> 414<ul> 415<li class="fragment"><span style="font-size: 85%;">This can be achieved by reducing a vector of features to a smaller set of scalesâe.g., via <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>principal component analysis</strong></span>âor partitioning the sample into a small number of unobserved groups (e.g., via <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>k</em>-means clustering</strong></span>).</span></li> 416</ul></li> 417</ul> 418</section> 419<section id="unsupervised-machine-learning-uml-cont." class="slide level2"> 420<h2>Unsupervised Machine Learning <span style="font-size: 70%;">(UML)</span> <em>cont</em>.</h2> 421<h3 id="an-illustration-1"><span style="color:#9c9c9c; font-size: 85%;">An Illustration</span></h3> 422<div class="cell"> 423<div class="cell-output-display"> 424<div id="tobweuylhp" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;"> 425<style>@import url("https://fonts.googleapis.com/css2?family=Chivo:ital,wght@0,100;0,200;0,300;0,400;0,500;0,600;0,700;0,800;0,900;1,100;1,200;1,300;1,400;1,500;1,600;1,700;1,800;1,900&display=swap"); 426@import url("https://fonts.googleapis.com/css2?family=Chivo:ital,wght@0,100;0,200;0,300;0,400;0,500;0,600;0,700;0,800;0,900;1,100;1,200;1,300;1,400;1,500;1,600;1,700;1,800;1,900&display=swap"); 427@import url("https://fonts.googleapis.com/css2?family=Chivo:ital,wght@0,100;0,200;0,300;0,400;0,500;0,600;0,700;0,800;0,900;1,100;1,200;1,300;1,400;1,500;1,600;1,700;1,800;1,900&display=swap"); 428@import url("https://fonts.googleapis.com/css2?family=Cairo:ital,wght@0,100;0,200;0,300;0,400;0,500;0,600;0,700;0,800;0,900;1,100;1,200;1,300;1,400;1,500;1,600;1,700;1,800;1,900&display=swap"); 429#tobweuylhp table { 430 font-family: Ubuntu; 431 -webkit-font-smoothing: antialiased; 432 -moz-osx-font-smoothing: grayscale; 433} 434 435#tobweuylhp thead, #tobweuylhp tbody, #tobweuylhp tfoot, #tobweuylhp tr, #tobweuylhp td, #tobweuylhp th { 436 border-style: none; 437} 438 439#tobweuylhp p { 440 margin: 0; 441 padding: 0; 442} 443 444#tobweuylhp .gt_table { 445 display: table; 446 border-collapse: collapse; 447 line-height: normal; 448 margin-left: auto; 449 margin-right: auto; 450 color: #333333; 451 font-size: 65%; 452 font-weight: 300; 453 font-style: normal; 454 background-color: #FFFFFF; 455 width: auto; 456 border-top-style: none; 457 border-top-width: 3px; 458 border-top-color: #A8A8A8; 459 border-right-style: none; 460 border-right-width: 2px; 461 border-right-color: #D3D3D3; 462 border-bottom-style: none; 463 border-bottom-width: 2px; 464 border-bottom-color: #A8A8A8; 465 border-left-style: none; 466 border-left-width: 2px; 467 border-left-color: #D3D3D3; 468} 469 470#tobweuylhp .gt_caption { 471 padding-top: 4px; 472 padding-bottom: 4px; 473} 474 475#tobweuylhp .gt_title { 476 color: #333333; 477 font-size: 125%; 478 font-weight: initial; 479 padding-top: 4px; 480 padding-bottom: 4px; 481 padding-left: 5px; 482 padding-right: 5px; 483 border-bottom-color: #FFFFFF; 484 border-bottom-width: 0; 485} 486 487#tobweuylhp .gt_subtitle { 488 color: #333333; 489 font-size: 85%; 490 font-weight: initial; 491 padding-top: 3px; 492 padding-bottom: 5px; 493 padding-left: 5px; 494 padding-right: 5px; 495 border-top-color: #FFFFFF; 496 border-top-width: 0; 497} 498 499#tobweuylhp .gt_heading { 500 background-color: #FFFFFF; 501 text-align: left; 502 border-bottom-color: #FFFFFF; 503 border-left-style: none; 504 border-left-width: 1px; 505 border-left-color: #D3D3D3; 506 border-right-style: none; 507 border-right-width: 1px; 508 border-right-color: #D3D3D3; 509} 510 511#tobweuylhp .gt_bottom_border { 512 border-bottom-style: none; 513 border-bottom-width: 2px; 514 border-bottom-color: #D3D3D3; 515} 516 517#tobweuylhp .gt_col_headings { 518 border-top-style: none; 519 border-top-width: 2px; 520 border-top-color: #D3D3D3; 521 border-bottom-style: solid; 522 border-bottom-width: 2px; 523 border-bottom-color: #000000; 524 border-left-style: none; 525 border-left-width: 1px; 526 border-left-color: #D3D3D3; 527 border-right-style: none; 528 border-right-width: 1px; 529 border-right-color: #D3D3D3; 530} 531 532#tobweuylhp .gt_col_heading { 533 color: #333333; 534 background-color: #FFFFFF; 535 font-size: 100%; 536 font-weight: normal; 537 text-transform: inherit; 538 border-left-style: none; 539 border-left-width: 1px; 540 border-left-color: #D3D3D3; 541 border-right-style: none; 542 border-right-width: 1px; 543 border-right-color: #D3D3D3; 544 vertical-align: bottom; 545 padding-top: 5px; 546 padding-bottom: 6px; 547 padding-left: 5px; 548 padding-right: 5px; 549 overflow-x: hidden; 550} 551 552#tobweuylhp .gt_column_spanner_outer {
553 color: #333333; 554 background-color: #FFFFFF; 555 font-size: 100%; 556 font-weight: normal; 557 text-transform: inherit; 558 padding-top: 0; 559 padding-bottom: 0; 560 padding-left: 4px; 561 padding-right: 4px; 562} 563 564#tobweuylhp .gt_column_spanner_outer:first-child { 565 padding-left: 0; 566} 567 568#tobweuylhp .gt_column_spanner_outer:last-child { 569 padding-right: 0; 570} 571 572#tobweuylhp .gt_column_spanner { 573 border-bottom-style: solid; 574 border-bottom-width: 2px; 575 border-bottom-color: #000000; 576 vertical-align: bottom; 577 padding-top: 5px; 578 padding-bottom: 5px; 579 overflow-x: hidden; 580 display: inline-block; 581 width: 100%; 582} 583 584#tobweuylhp .gt_spanner_row { 585 border-bottom-style: hidden; 586} 587 588#tobweuylhp .gt_group_heading { 589 padding-top: 8px; 590 padding-bottom: 8px; 591 padding-left: 5px; 592 padding-right: 5px; 593 color: #333333; 594 background-color: #FFFFFF; 595 font-size: 100%; 596 font-weight: initial; 597 text-transform: inherit; 598 border-top-style: none; 599 border-top-width: 2px; 600 border-top-color: #000000; 601 border-bottom-style: solid; 602 border-bottom-width: 1px; 603 border-bottom-color: #FFFFFF; 604 border-left-style: none; 605 border-left-width: 1px; 606 border-left-color: #D3D3D3; 607 border-right-style: none; 608 border-right-width: 1px; 609 border-right-color: #D3D3D3; 610 vertical-align: middle; 611 text-align: left; 612} 613 614#tobweuylhp .gt_empty_group_heading { 615 padding: 0.5px; 616 color: #333333; 617 background-color: #FFFFFF; 618 font-size: 100%; 619 font-weight: initial; 620 border-top-style: none; 621 border-top-width: 2px; 622 border-top-color: #000000; 623 border-bottom-style: solid; 624 border-bottom-width: 1px; 625 border-bottom-color: #FFFFFF; 626 vertical-align: middle; 627} 628 629#tobweuylhp .gt_from_md > :first-child { 630 margin-top: 0; 631} 632 633#tobweuylhp .gt_from_md > :last-child { 634 margin-bottom: 0; 635} 636 637#tobweuylhp .gt_row { 638 padding-top: 3px; 639 padding-bottom: 3px; 640 padding-left: 5px; 641 padding-right: 5px; 642 margin: 10px; 643 border-top-style: solid; 644 border-top-width: 1px; 645 border-top-color: #D3D3D3; 646 border-left-style: none; 647 border-left-width: 1px; 648 border-left-color: #D3D3D3; 649 border-right-style: none; 650 border-right-width: 1px; 651 border-right-color: #D3D3D3; 652 vertical-align: middle; 653 overflow-x: hidden; 654} 655 656#tobweuylhp .gt_stub { 657 color: #333333; 658 background-color: #FFFFFF; 659 font-size: 100%; 660 font-weight: initial; 661 text-transform: inherit; 662 border-right-style: solid; 663 border-right-width: 0px; 664 border-right-color: #FFFFFF; 665 padding-left: 5px; 666 padding-right: 5px; 667} 668 669#tobweuylhp .gt_stub_row_group { 670 color: #333333; 671 background-color: #FFFFFF; 672 font-size: 100%; 673 font-weight: initial; 674 text-transform: inherit; 675 border-right-style: solid; 676 border-right-width: 2px; 677 border-right-color: #D3D3D3; 678 padding-left: 5px; 679 padding-right: 5px; 680 vertical-align: top; 681} 682 683#tobweuylhp .gt_row_group_first td { 684 border-top-width: 2px; 685} 686 687#tobweuylhp .gt_row_group_first th { 688 border-top-width: 2px; 689} 690 691#tobweuylhp .gt_summary_row { 692 color: #333333; 693 background-color: #FFFFFF; 694 text-transform: inherit; 695 padding-top: 8px; 696 padding-bottom: 8px; 697 padding-left: 5px; 698 padding-right: 5px; 699} 700 701#tobweuylhp .gt_first_summary_row { 702 border-top-style: solid; 703 border-top-color: #D3D3D3; 704} 705 706#tobweuylhp .gt_first_summary_row.thick { 707 border-top-width: 2px; 708} 709 710#tobweuylhp .gt_last_summary_row { 711 padding-top: 8px; 712 padding-bottom: 8px; 713 padding-left: 5px; 714 padding-right: 5px; 715 border-bottom-style: solid; 716 border-bottom-width: 2px; 717 border-bottom-color: #D3D3D3; 718} 719 720#tobweuylhp .gt_grand_summary_row { 721 color: #333333; 722 background-color: #FFFFFF; 723 text-transform: inherit; 724 padding-top: 8px; 725 padding-bottom: 8px; 726 padding-left: 5px; 727 padding-right: 5px; 728} 729 730#tobweuylhp .gt_first_grand_summary_row { 731 padding-top: 8px; 732 padding-bottom: 8px; 733 padding-left: 5px; 734 padding-right: 5px; 735 border-top-style: double; 736 border-top-width: 6px; 737 border-top-color: #D3D3D3; 738} 739 740#tobweuylhp .gt_last_grand_summary_row_top { 741 padding-top: 8px; 742 padding-bottom: 8px; 743 padding-left: 5px; 744 padding-right: 5px; 745 border-bottom-style: double; 746 border-bottom-width: 6px; 747 border-bottom-color: #D3D3D3; 748} 749 750#tobweuylhp .gt_striped { 751 background-color: rgba(128, 128, 128, 0.05); 752} 753 754#tobweuylhp .gt_table_body { 755 border-top-style: solid; 756 border-top-width: 2px; 757 border-top-color: #D3D3D3; 758 border-bottom-style: solid; 759 border-bottom-width: 2px; 760 border-bottom-color: #D3D3D3; 761} 762 763#tobweuylhp .gt_footnotes { 764 color: #333333; 765 background-color: #FFFFFF; 766 border-bottom-style: none; 767 border-bottom-width: 2px; 768 border-bottom-color: #D3D3D3; 769 border-left-style: none; 770 border-left-width: 2px; 771 border-left-color: #D3D3D3; 772 border-right-style: none; 773 border-right-width: 2px; 774 border-right-color: #D3D3D3; 775} 776 777#tobweuylhp .gt_footnote { 778 margin: 0px; 779 font-size: 90%; 780 padding-top: 4px; 781 padding-bottom: 4px; 782 padding-left: 5px; 783 padding-right: 5px; 784} 785 786#tobweuylhp .gt_sourcenotes { 787 color: #333333; 788 background-color: #FFFFFF; 789 border-bottom-style: none; 790 border-bottom-width: 2px; 791 border-bottom-color: #D3D3D3; 792 border-left-style: none; 793 border-left-width: 2px; 794 border-left-color: #D3D3D3; 795 border-right-style: none; 796 border-right-width: 2px; 797 border-right-color: #D3D3D3; 798} 799 800#tobweuylhp .gt_sourcenote { 801 font-size: 12px; 802 padding-top: 4px; 803 padding-bottom: 4px; 804 padding-left: 5px; 805 padding-right: 5px; 806} 807 808#tobweuylhp .gt_left { 809 text-align: left; 810} 811 812#tobweuylhp .gt_center { 813 text-align: center; 814} 815 816#tobweuylhp .gt_right { 817 text-align: right; 818 font-variant-numeric: tabular-nums; 819} 820 821#tobweuylhp .gt_font_normal { 822 font-weight: normal; 823} 824 825#tobweuylhp .gt_font_bold { 826 font-weight: bold; 827} 828 829#tobweuylhp .gt_font_italic { 830 font-style: italic; 831} 832 833#tobweuylhp .gt_super { 834 font-size: 65%; 835} 836 837#tobweuylhp .gt_footnote_marks { 838 font-size: 75%; 839 vertical-align: 0.4em; 840 position: initial; 841} 842 843#tobweuylhp .gt_asterisk { 844 font-size: 100%; 845 vertical-align: 0; 846} 847 848#tobweuylhp .gt_indent_1 { 849 text-indent: 5px; 850} 851 852#tobweuylhp .gt_indent_2 { 853 text-indent: 10px; 854} 855 856#tobweuylhp .gt_indent_3 { 857 text-indent: 15px; 858} 859 860#tobweuylhp .gt_indent_4 { 861 text-indent: 20px; 862} 863 864#tobweuylhp .gt_indent_5 { 865 text-indent: 25px; 866} 867 868#tobweuylhp .katex-display { 869 display: inline-flex !important; 870 margin-bottom: 0.75em !important; 871} 872 873#tobweuylhp div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { 874 height: 0px !important; 875} 876 877tbody tr:last-child { 878 border-bottom: 2px solid #ffffff00; 879} 880</style> 881 882<table class="gt_table caption-top" data-quarto-postprocess="true" data-quarto-disable-processing="false" data-quarto-bootstrap="false"> 883<thead> 884<tr class="gt_col_headings header"> 885<th id="V-Party-Name" class="gt_col_heading gt_columns_bottom_border gt_center" data-quarto-table-cell-role="th" style="border-top-width: 0px; border-top-style: solid; border-top-color: black; font-family: Chivo; font-size: 14px; vertical-align: bottom; font-weight: 200; text-transform: uppercase" scope="col">V-Party Name</th> 886<th id="Label" class="gt_col_heading gt_columns_bottom_border gt_center" data-quarto-table-cell-role="th" style="border-top-width: 0px; border-top-style: solid; border-top-color: black; font-family: Chivo; font-size: 14px; vertical-align: bottom; font-weight: 200; text-transform: uppercase" scope="col">Label</th> 887<th id="Underlying-Question" class="gt_col_heading gt_columns_bottom_border gt_left" data-quarto-table-cell-role="th" style="border-top-width: 0px; border-top-style: solid; border-top-color: black; font-family: Chivo; font-size: 14px; vertical-align: bottom; font-weight: 200; text-transform: uppercase" scope="col">Underlying Question</th> 888</tr> 889</thead> 890<tbody class="gt_table_body"> 891<tr class="odd"> 892<td class="gt_row gt_center" headers="V-Party Name"><code>v2paanteli</code></td> 893<td class="gt_row gt_center" headers="Label">Anti-Elitism</td> 894<td class="gt_row gt_left" headers="Underlying Question">How important is anti-elite rhetoric for this party?</td> 895</tr> 896<tr class="even"> 897<td class="gt_row gt_center" headers="V-Party Name"><code>v2papeople</code></td> 898<td class="gt_row gt_center" headers="Label">People-Centrism</td> 899<td class="gt_row gt_left" headers="Underlying Question">Do leaders of this party glorify the ordinary people and identify themselves as part of them?</td> 900</tr> 901<tr class="odd"> 902<td class="gt_row gt_center" headers="V-Party Name"><code>v2paculsup</code></td> 903<td class="gt_row gt_center" headers="Label">Cultural Chauvinism</td> 904<td class="gt_row gt_left" headers="Underlying Question">To what extent does the party leadership promote the cultural superiority of a specific social group or the nation as a whole?</td> 905</tr> 906<tr class="even"> 907<td class="gt_row gt_center" headers="V-Party Name"><code>v2paminor</code></td> 908<td class="gt_row gt_center" headers="Label">Minority Rights</td> 909<td class="gt_row gt_left" headers="Underlying Question">According to the leadership of this party, how often should the will of the majority be implemented even if doing so w
909ould violate the rights of minorities?</td> 910</tr> 911<tr class="odd"> 912<td class="gt_row gt_center" headers="V-Party Name"><code>v2paplur</code></td> 913<td class="gt_row gt_center" headers="Label">Political Pluralism</td> 914<td class="gt_row gt_left" headers="Underlying Question">Prior to this election, to what extent was the leadership of this political party clearly committed to free and fair elections with multiple parties, freedom of speech, media, assembly and association?</td> 915</tr> 916<tr class="even"> 917<td class="gt_row gt_center" headers="V-Party Name"><code>v2paopresp</code></td> 918<td class="gt_row gt_center" headers="Label">Demonization of Opponents</td> 919<td class="gt_row gt_left" headers="Underlying Question">Prior to this election, have leaders of this party used severe personal attacks or tactics of demonization against their opponents?</td> 920</tr> 921<tr class="odd"> 922<td class="gt_row gt_center" headers="V-Party Name"><code>v2paviol</code></td> 923<td class="gt_row gt_center" headers="Label">Rejection of Political Violence</td> 924<td class="gt_row gt_left" headers="Underlying Question">To what extent does the leadership of this party explicitly discourage the use of violence against domestic political opponents?</td> 925</tr> 926</tbody> 927</table> 928 929</div> 930</div> 931</div> 932</section> 933<section id="unsupervised-machine-learning-uml-cont.-1" class="slide level2 smaller"> 934<h2>Unsupervised Machine Learning <span style="font-size: 70%;">(UML)</span> <em>cont</em>.</h2> 935<h3 id="an-illustration-2"><span style="color:#9c9c9c; font-size: 85%;">An Illustration</span></h3> 936 937<img data-src="./media/pca_shift.gif" class="r-stretch quarto-figure-center"><p class="caption"><span style="font-size: 65%; text-align: center; margin-top: -10px">Karim and Lukkâs <em>The Radicalization of Mainstream Parties in the 21<span style="font-size: 65%;"><sup>st</sup></span> Century</em></span></p></section></section> 938<section> 939<section id="machine-learning-vs-classical-statistics" class="title-slide slide level1 center" data-background-color="#43464B"> 940<h1>Machine Learning vs Classical Statistics</h1> 941 942</section> 943<section id="the-two-cultures" class="slide level2"> 944<h2>The Two Cultures</h2> 945<div style="margin-top: 20px; font-size: 85%;"> 946<ul> 947<li class="fragment"><p>As Grimmer and colleagues <span class="citation" data-cites="grimmer2021">(<a href="#/references" role="doc-biblioref" onclick="">2021</a>)</span> note, âmachine learning is as much a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>culture</strong></span> defined by a distinct set of values and tools as it is a set of algorithms.â</p></li> 948<li class="fragment"><p>This point has, of course, been made elsewhere.</p> 949<ul> 950<li class="fragment"><p><span style="font-size: 95%;">Leo Breiman <span class="citation" data-cites="breiman2001">(<a href="#/references" role="doc-biblioref" onclick="">2001</a>)</span> famously used the imagery of warring <em>cultures</em> to describe two major traditionsâ(i) the <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>generative modelling culture</strong></span> and (ii) the <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>predictive modelling culture</strong></span>âthat have achieved hegemony within the world of statistical modelling.</span></p></li> 951<li class="fragment"><p><span style="font-size: 95%;">The terms <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>generative</em></span> and <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>predictive</em></span> (as opposed to <em>data</em> and <em>algorithmic</em>) come from David Donohoâs <span class="citation" data-cites="donoho2017">(<a href="#/references" role="doc-biblioref" onclick="">2017</a>)</span> <em>50 Years of Data Science</em>.</span></p></li> 952</ul></li> 953</ul> 954</div> 955</section> 956<section id="the-two-cultures-cont." class="slide level2 no-hover"> 957<h2>The Two Cultures <em>cont</em>.</h2> 958<p><br></p> 959<div class="cell"> 960<div class="cell-output-display"> 961<div id="ofqqljgkfa" style="padding-left:0px;padding-right:0px;padding-top:10px;padding-bottom:10px;overflow-x:auto;overflow-y:auto;width:auto;height:auto;"> 962<style>@import url("https://fonts.googleapis.com/css2?family=Chivo:ital,wght@0,100;0,200;0,300;0,400;0,500;0,600;0,700;0,800;0,900;1,100;1,200;1,300;1,400;1,500;1,600;1,700;1,800;1,900&display=swap"); 963@import url("https://fonts.googleapis.com/css2?family=Chivo:ital,wght@0,100;0,200;0,300;0,400;0,500;0,600;0,700;0,800;0,900;1,100;1,200;1,300;1,400;1,500;1,600;1,700;1,800;1,900&display=swap"); 964@import url("https://fonts.googleapis.com/css2?family=Chivo:ital,wght@0,100;0,200;0,300;0,400;0,500;0,600;0,700;0,800;0,900;1,100;1,200;1,300;1,400;1,500;1,600;1,700;1,800;1,900&display=swap"); 965@import url("https://fonts.googleapis.com/css2?family=Cairo:ital,wght@0,100;0,200;0,300;0,400;0,500;0,600;0,700;0,800;0,900;1,100;1,200;1,300;1,400;1,500;1,600;1,700;1,800;1,900&display=swap"); 966#ofqqljgkfa table { 967 font-family: Ubuntu; 968 -webkit-font-smoothing: antialiased; 969 -moz-osx-font-smoothing: grayscale; 970} 971 972#ofqqljgkfa thead, #ofqqljgkfa tbody, #ofqqljgkfa tfoot, #ofqqljgkfa tr, #ofqqljgkfa td, #ofqqljgkfa th { 973 border-style: none; 974} 975 976#ofqqljgkfa p { 977 margin: 0; 978 padding: 0; 979} 980 981#ofqqljgkfa .gt_table { 982 display: table; 983 border-collapse: collapse; 984 line-height: normal; 985 margin-left: auto; 986 margin-right: auto; 987 color: #333333; 988 font-size: 63%; 989 font-weight: 300; 990 font-style: normal; 991 background-color: #FFFFFF; 992 width: auto; 993 border-top-style: none; 994 border-top-width: 3px; 995 border-top-color: #A8A8A8; 996 border-right-style: none; 997 border-right-width: 2px; 998 border-right-color: #D3D3D3; 999 border-bottom-style: none; 1000 border-bottom-width: 2px; 1001 border-bottom-color: #A8A8A8; 1002 border-left-style: none; 1003 border-left-width: 2px; 1004 border-left-color: #D3D3D3; 1005} 1006 1007#ofqqljgkfa .gt_caption { 1008 padding-top: 4px; 1009 padding-bottom: 4px; 1010} 1011 1012#ofqqljgkfa .gt_title { 1013 color: #333333; 1014 font-size: 125%; 1015 font-weight: initial; 1016 padding-top: 4px; 1017 padding-bottom: 4px; 1018 padding-left: 5px; 1019 padding-right: 5px; 1020 border-bottom-color: #FFFFFF; 1021 border-bottom-width: 0; 1022} 1023 1024#ofqqljgkfa .gt_subtitle { 1025 color: #333333; 1026 font-size: 85%; 1027 font-weight: initial; 1028 padding-top: 3px; 1029 padding-bottom: 5px; 1030 padding-left: 5px; 1031 padding-right: 5px; 1032 border-top-color: #FFFFFF; 1033 border-top-width: 0; 1034} 1035 1036#ofqqljgkfa .gt_heading { 1037 background-color: #FFFFFF; 1038 text-align: left; 1039 border-bottom-color: #FFFFFF; 1040 border-left-style: none; 1041 border-left-width: 1px; 1042 border-left-color: #D3D3D3; 1043 border-right-style: none; 1044 border-right-width: 1px; 1045 border-right-color: #D3D3D3; 1046} 1047 1048#ofqqljgkfa .gt_bottom_border { 1049 border-bottom-style: none; 1050 border-bottom-width: 2px; 1051 border-bottom-color: #D3D3D3; 1052} 1053 1054#ofqqljgkfa .gt_col_headings { 1055 border-top-style: none; 1056 border-top-width: 2px; 1057 border-top-color: #D3D3D3; 1058 border-bottom-style: solid; 1059 border-bottom-width: 2px; 1060 border-bottom-color: #000000; 1061 border-left-style: none; 1062 border-left-width: 1px; 1063 border-left-color: #D3D3D3; 1064 border-right-style: none; 1065 border-right-width: 1px; 1066 border-right-color: #D3D3D3; 1067} 1068 1069#ofqqljgkfa .gt_col_heading { 1070 color: #333333; 1071 background-color: #FFFFFF; 1072 font-size: 100%; 1073 font-weight: normal; 1074 text-transform: inherit; 1075 border-left-style: none; 1076 border-left-width: 1px; 1077 border-left-color: #D3D3D3; 1078 border-right-style: none; 1079 border-right-width: 1px; 1080 border-right-color: #D3D3D3; 1081 vertical-align: bottom; 1082 padding-top: 5px; 1083 padding-bottom: 6px; 1084 padding-left: 10px; 1085 padding-right: 10px; 1086 overflow-x: hidden; 1087} 1088 1089#ofqqljgkfa .gt_column_spanner_outer {
1090 color: #333333; 1091 background-color: #FFFFFF; 1092 font-size: 100%; 1093 font-weight: normal; 1094 text-transform: inherit; 1095 padding-top: 0; 1096 padding-bottom: 0; 1097 padding-left: 4px; 1098 padding-right: 4px; 1099} 1100 1101#ofqqljgkfa .gt_column_spanner_outer:first-child { 1102 padding-left: 0; 1103} 1104 1105#ofqqljgkfa .gt_column_spanner_outer:last-child { 1106 padding-right: 0; 1107} 1108 1109#ofqqljgkfa .gt_column_spanner { 1110 border-bottom-style: solid; 1111 border-bottom-width: 2px; 1112 border-bottom-color: #000000; 1113 vertical-align: bottom; 1114 padding-top: 5px; 1115 padding-bottom: 5px; 1116 overflow-x: hidden; 1117 display: inline-block; 1118 width: 100%; 1119} 1120 1121#ofqqljgkfa .gt_spanner_row { 1122 border-bottom-style: hidden; 1123} 1124 1125#ofqqljgkfa .gt_group_heading { 1126 padding-top: 8px; 1127 padding-bottom: 8px; 1128 padding-left: 5px; 1129 padding-right: 5px; 1130 color: #FFFFFF; 1131 background-color: #5D3C85; 1132 font-size: 100%; 1133 font-weight: initial; 1134 text-transform: inherit; 1135 border-top-style: none; 1136 border-top-width: 2px; 1137 border-top-color: #000000; 1138 border-bottom-style: solid; 1139 border-bottom-width: 1px; 1140 border-bottom-color: #FFFFFF; 1141 border-left-style: none; 1142 border-left-width: 1px; 1143 border-left-color: #D3D3D3; 1144 border-right-style: none; 1145 border-right-width: 1px; 1146 border-right-color: #D3D3D3; 1147 vertical-align: middle; 1148 text-align: left; 1149} 1150 1151#ofqqljgkfa .gt_empty_group_heading { 1152 padding: 0.5px; 1153 color: #FFFFFF; 1154 background-color: #5D3C85; 1155 font-size: 100%; 1156 font-weight: initial; 1157 border-top-style: none; 1158 border-top-width: 2px; 1159 border-top-color: #000000; 1160 border-bottom-style: solid; 1161 border-bottom-width: 1px; 1162 border-bottom-color: #FFFFFF; 1163 vertical-align: middle; 1164} 1165 1166#ofqqljgkfa .gt_from_md > :first-child { 1167 margin-top: 0; 1168} 1169 1170#ofqqljgkfa .gt_from_md > :last-child { 1171 margin-bottom: 0; 1172} 1173 1174#ofqqljgkfa .gt_row { 1175 padding-top: 3px; 1176 padding-bottom: 3px; 1177 padding-left: 5px; 1178 padding-right: 5px; 1179 margin: 10px; 1180 border-top-style: solid; 1181 border-top-width: 1px; 1182 border-top-color: #D3D3D3; 1183 border-left-style: none; 1184 border-left-width: 1px; 1185 border-left-color: #D3D3D3; 1186 border-right-style: none; 1187 border-right-width: 1px; 1188 border-right-color: #D3D3D3; 1189 vertical-align: middle; 1190 overflow-x: hidden; 1191} 1192 1193#ofqqljgkfa .gt_stub { 1194 color: #333333; 1195 background-color: #FFFFFF; 1196 font-size: 100%; 1197 font-weight: initial; 1198 text-transform: inherit; 1199 border-right-style: solid; 1200 border-right-width: 0px; 1201 border-right-color: #FFFFFF; 1202 padding-left: 5px; 1203 padding-right: 5px; 1204} 1205 1206#ofqqljgkfa .gt_stub_row_group { 1207 color: #333333; 1208 background-color: #FFFFFF; 1209 font-size: 100%; 1210 font-weight: initial; 1211 text-transform: inherit; 1212 border-right-style: solid; 1213 border-right-width: 2px; 1214 border-right-color: #D3D3D3; 1215 padding-left: 5px; 1216 padding-right: 5px; 1217 vertical-align: top; 1218} 1219 1220#ofqqljgkfa .gt_row_group_first td { 1221 border-top-width: 2px; 1222} 1223 1224#ofqqljgkfa .gt_row_group_first th { 1225 border-top-width: 2px; 1226} 1227 1228#ofqqljgkfa .gt_summary_row { 1229 color: #333333; 1230 background-color: #FFFFFF; 1231 text-transform: inherit; 1232 padding-top: 8px; 1233 padding-bottom: 8px; 1234 padding-left: 5px; 1235 padding-right: 5px; 1236} 1237 1238#ofqqljgkfa .gt_first_summary_row { 1239 border-top-style: solid; 1240 border-top-color: #D3D3D3; 1241} 1242 1243#ofqqljgkfa .gt_first_summary_row.thick { 1244 border-top-width: 2px; 1245} 1246 1247#ofqqljgkfa .gt_last_summary_row { 1248 padding-top: 8px; 1249 padding-bottom: 8px; 1250 padding-left: 5px; 1251 padding-right: 5px; 1252 border-bottom-style: solid; 1253 border-bottom-width: 2px; 1254 border-bottom-color: #D3D3D3; 1255} 1256 1257#ofqqljgkfa .gt_grand_summary_row { 1258 color: #333333; 1259 background-color: #FFFFFF; 1260 text-transform: inherit; 1261 padding-top: 8px; 1262 padding-bottom: 8px; 1263 padding-left: 5px; 1264 padding-right: 5px; 1265} 1266 1267#ofqqljgkfa .gt_first_grand_summary_row { 1268 padding-top: 8px; 1269 padding-bottom: 8px; 1270 padding-left: 5px; 1271 padding-right: 5px; 1272 border-top-style: double; 1273 border-top-width: 6px; 1274 border-top-color: #D3D3D3; 1275} 1276 1277#ofqqljgkfa .gt_last_grand_summary_row_top { 1278 padding-top: 8px; 1279 padding-bottom: 8px; 1280 padding-left: 5px; 1281 padding-right: 5px; 1282 border-bottom-style: double; 1283 border-bottom-width: 6px; 1284 border-bottom-color: #D3D3D3; 1285} 1286 1287#ofqqljgkfa .gt_striped { 1288 background-color: rgba(128, 128, 128, 0.05); 1289} 1290 1291#ofqqljgkfa .gt_table_body { 1292 border-top-style: solid; 1293 border-top-width: 2px; 1294 border-top-color: #D3D3D3; 1295 border-bottom-style: solid; 1296 border-bottom-width: 2px; 1297 border-bottom-color: #D3D3D3; 1298} 1299 1300#ofqqljgkfa .gt_footnotes { 1301 color: #333333; 1302 background-color: #FFFFFF; 1303 border-bottom-style: none; 1304 border-bottom-width: 2px; 1305 border-bottom-color: #D3D3D3; 1306 border-left-style: none; 1307 border-left-width: 2px; 1308 border-left-color: #D3D3D3; 1309 border-right-style: none; 1310 border-right-width: 2px; 1311 border-right-color: #D3D3D3; 1312} 1313 1314#ofqqljgkfa .gt_footnote { 1315 margin: 0px; 1316 font-size: 90%; 1317 padding-top: 4px; 1318 padding-bottom: 4px; 1319 padding-left: 5px; 1320 padding-right: 5px; 1321} 1322 1323#ofqqljgkfa .gt_sourcenotes { 1324 color: #333333; 1325 background-color: #FFFFFF; 1326 border-bottom-style: none; 1327 border-bottom-width: 2px; 1328 border-bottom-color: #D3D3D3; 1329 border-left-style: none; 1330 border-left-width: 2px; 1331 border-left-color: #D3D3D3; 1332 border-right-style: none; 1333 border-right-width: 2px; 1334 border-right-color: #D3D3D3; 1335} 1336 1337#ofqqljgkfa .gt_sourcenote { 1338 font-size: 12px; 1339 padding-top: 4px; 1340 padding-bottom: 4px; 1341 padding-left: 5px; 1342 padding-right: 5px; 1343} 1344 1345#ofqqljgkfa .gt_left { 1346 text-align: left; 1347} 1348 1349#ofqqljgkfa .gt_center { 1350 text-align: center; 1351} 1352 1353#ofqqljgkfa .gt_right { 1354 text-align: right; 1355 font-variant-numeric: tabular-nums; 1356} 1357 1358#ofqqljgkfa .gt_font_normal { 1359 font-weight: normal; 1360} 1361 1362#ofqqljgkfa .gt_font_bold { 1363 font-weight: bold; 1364} 1365 1366#ofqqljgkfa .gt_font_italic { 1367 font-style: italic; 1368} 1369 1370#ofqqljgkfa .gt_super { 1371 font-size: 65%; 1372} 1373 1374#ofqqljgkfa .gt_footnote_marks { 1375 font-size: 75%; 1376 vertical-align: 0.4em; 1377 position: initial; 1378} 1379 1380#ofqqljgkfa .gt_asterisk { 1381 font-size: 100%; 1382 vertical-align: 0; 1383} 1384 1385#ofqqljgkfa .gt_indent_1 { 1386 text-indent: 5px; 1387} 1388 1389#ofqqljgkfa .gt_indent_2 { 1390 text-indent: 10px; 1391} 1392 1393#ofqqljgkfa .gt_indent_3 { 1394 text-indent: 15px; 1395} 1396 1397#ofqqljgkfa .gt_indent_4 { 1398 text-indent: 20px; 1399} 1400 1401#ofqqljgkfa .gt_indent_5 { 1402 text-indent: 25px; 1403} 1404 1405#ofqqljgkfa .katex-display { 1406 display: inline-flex !important; 1407 margin-bottom: 0.75em !important; 1408} 1409 1410#ofqqljgkfa div.Reactable > div.rt-table > div.rt-thead > div.rt-tr.rt-tr-group-header > div.rt-th-group:after { 1411 height: 0px !important; 1412} 1413 1414tbody tr:last-child { 1415 border-bottom: 2px solid #ffffff00; 1416} 1417</style> 1418 1419<table class="gt_table caption-top" data-quarto-postprocess="true" data-quarto-disable-processing="false" data-quarto-bootstrap="false"> 1420<thead> 1421<tr class="gt_col_headings header"> 1422<th id="Quantity-of-Interest" class="gt_col_heading gt_columns_bottom_border gt_center" data-quarto-table-cell-role="th" style="border-top-width: 0px; border-top-style: solid; border-top-color: black; font-family: Chivo; font-size: 14px; vertical-align: bottom; font-weight: 200; text-transform: uppercase" scope="col">Quantity of Interest</th> 1423<th id="Primary-Goals" class="gt_col_heading gt_columns_bottom_border gt_left" data-quarto-table-cell-role="th" style="border-top-width: 0px; border-top-style: solid; border-top-color: black; font-family: Chivo; font-size: 14px; vertical-align: bottom; font-weight: 200; text-transform: uppercase" scope="col">Primary Goals</th> 1424<th id="Key-Strengths" class="gt_col_heading gt_columns_bottom_border gt_left" data-quarto-table-cell-role="th" style="border-top-width: 0px; border-top-style: solid; border-top-color: black; font-family: Chivo; font-size: 14px; vertical-align: bottom; font-weight: 200; text-transform: uppercase" scope="col">Key Strengths</th> 1425<th id="Key-Limitations" class="gt_col_heading gt_columns_bottom_border gt_left" data-quarto-table-cell-role="th" style="border-top-width: 0px; border-top-style: solid; border-top-color: black; font-family: Chivo; font-size: 14px; vertical-align: bottom; font-weight: 200; text-transform: uppercase" scope="col">Key Limitations</th> 1426</tr> 1427</thead> 1428<tbody class="gt_table_body"> 1429<tr class="gt_group_heading_row odd"> 1430<td colspan="4" id="Generative (i.e., Classical Statistics)" class="gt_group_heading" data-quarto-table-cell-role="th" style="border-bottom-width: 1px; border-bottom-style: solid; border-bottom-color: black" scope="colgroup">Generative (i.e., Classical Statistics)</td> 1431</tr> 1432<tr class="gt_row_group_first even"> 1433<td class="gt_row gt_center" headers="Generative (i.e., Classical Statistics) Quantity of Interest"><img data-src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAKYAAAD6CAYAAAArxDz7AAAMP2lDQ1BJQ0MgUHJvZmlsZQAASImVVwdYU8kWnluSkEBoAQSkhN4EkRpASggt9N5EJSQBQokxEFTs6KKCaxcL2NBVEQUrzYIidhbF3hcLKsq6WLArb1JA133le/N9c+e//5z5z5lz5965A4DacY5IlIeqA5AvLBTHBvvTk1NS6aSnAAcUQAVeQJnDLRAxo6PDASxD7d/Lu+sAkbZX7KVa/+z/r0WDxy/gAoBEQ5zBK+DmQ3wQALyKKxIXAkCU8mZTCkVSDCvQEsMAIV4oxVlyXCXFGXK8V2YTH8uCuB0AJRUOR5wFgOolyNOLuFlQQ7UfYkchTyAEQ
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
1433pF3ZM9J6+e3C8OE28vQArGFzQVasEf+Q6JteQXyYMin6LKog9TEqccmKoxVTi1Y5rttEXTnhYHFf82HZ/Ond42w2TG3BkPZjJnbpmFzMqY1TbbbPb82T1zgufsnEuZmzv39xLHkhUlb+clzWuZbzh/zvxHvwT/UluqWiouvbHAa8GmhfhCwcLORc6L1i36VsYrO1/uWF5R/mUxd/H5X8f8uvbXwSWZSzqXui3duIy4TLjs+nLf5TtXaKwoXvFoZcTKhlX0VWWr3q6euPpchUvFpjWUNZI13WvD1zatM1+3bN2X9dnrr1X6V9ZXGVQtqnq/gbfh8ka/jXWbDDeVb/q0WbD55pbgLQ3VltUVW4lbi7Y+2Za47cxvjN9qtutvL9/+dYdwR/fO2J3tNe41NbsMdi2tRWsltb2703Zf2hOwp6nOvm5LvU59+V6wV7L3+b70fdf3h+1vO8A4UHfQ4mDVIdqhsgakYVpDf2N2Y3dTSlNXc2hzW4tXy6HDDod3HDE5UnlU++jSY5Rj848Ntha3DhwXHe87kXXiUdvEtjsnk09ebY9p7zwVdurs6aDTJ88wz7Se9T575JznuebzjPONF9wuNHS4dhz63fX3Q51unQ0X3S82XfK41NI1tuvYZd/LJ64EXDl9lX31wrXIa13XE67fvJF2o/sm7+azW3m3Xt0uuv35zpy7hLtl99TvVdw3uF/9h80f9d1u3UcfBDzoeBj38M4j7qMXjwsef+mZ/4T6pOKp8dOaZ07PjvQG9V56Pu55zwvRi899pX9q/Fn10vrlwb/8/uroT+7veSV+Nfh68Ru9NzveurxtG4geuP8u/93n92Uf9D7s/Mj4eOZT0qenn6d8IX1Z+9Xma8u3sG93B/MHB0UcMUf2K4DBimZmAvB6BwDUFABo8HxGGSc//8kKIj+zyhD4T1h+RpQVNwDq4P97TB/8u7kBwN5t8PgF9dXSAIimAhDvAVBn5+E6dFaTnSulhQjPAZtDvmbkZ4B/U+Rnzh/i/rkFUlUX8HP7L10DfGN2EiXzAAAAimVYSWZNTQAqAAAACAAEARoABQAAAAEAAAA+ARsABQAAAAEAAABGASgAAwAAAAEAAgAAh2kABAAAAAEAAABOAAAAAAAAAJAAAAABAAAAkAAAAAEAA5KGAAcAAAASAAAAeKACAAQAAAABAAAApqADAAQAAAABAAAA+gAAAABBU0NJSQAAAFNjcmVlbnNob3R86jOHAAAACXBIWXMAABYlAAAWJQFJUiTwAAAB1mlUWHRYTUw6Y29tLmFkb2JlLnhtcAAAAAAAPHg6eG1wbWV0YSB4bWxuczp4PSJhZG9iZTpuczptZXRhLyIgeDp4bXB0az0iWE1QIENvcmUgNi4wLjAiPgogICA8cmRmOlJERiB4bWxuczpyZGY9Imh0dHA6Ly93d3cudzMub3JnLzE5OTkvMDIvMjItcmRmLXN5bnRheC1ucyMiPgogICAgICA8cmRmOkRlc2NyaXB0aW9uIHJkZjphYm91dD0iIgogICAgICAgICAgICB4bWxuczpleGlmPSJodHRwOi8vbnMuYWRvYmUuY29tL2V4aWYvMS4wLyI+CiAgICAgICAgIDxleGlmOlBpeGVsWURpbWVuc2lvbj4yNTA8L2V4aWY6UGl4ZWxZRGltZW5zaW9uPgogICAgICAgICA8ZXhpZjpQaXhlbFhEaW1lbnNpb24+MTY2PC9leGlmOlBpeGVsWERpbWVuc2lvbj4KICAgICAgICAgPGV4aWY6VXNlckNvbW1lbnQ+U2NyZWVuc2hvdDwvZXhpZjpVc2VyQ29tbWVudD4KICAgICAgPC9yZGY6RGVzY3JpcHRpb24+CiAgIDwvcmRmOlJERj4KPC94OnhtcG1ldGE+CoHC6M0AAAAcaURPVAAAAAIAAAAAAAAAfQAAACgAAAB9AAAAfQAAK7NpscXXAAArf0lEQVR4Aex9ebRnVXXmfcUkoCDFVDKlO44xQAScQGhdGsW1nJAoJkKBLYgDf7Rtora9Ii0OAaLEMUFRC1y2YxtxOUGMqA0miEMsQEAlXQjiEg1VBUoxV93e++zz7fPtc+791atX9apewe+u9e7+9t7f2fucfc/v3OE3vJletm5gU+MM28XQiwE2bTWTFeNSC8Ak+66XVmhnIUHgBBE3DDFM8z946j+DidlMH555cc4kreEHDqYVZHZChZQpi1I38ab5yys/1NaUpl6Bg7pCZidUyAVc/5l+ncyAuJxRd2WVUndeGn08oQgoFKZYVYRhNVlTvIGgbJrmf5DWHyvm8PQKU0QoE2ZwnoAylRPP93z+BWd0NeR8SmadcQ40YEIbc8l+mr85bLLajKzGdUFZZzz/9U+ncl6V7MBqYr6AtE7pXjdMz2KNFmPRPjWUHS5KyaVwmr+sitP62yyza8w0w9LO1ijzlRlYTaQhNTQJSstu3GqYSbtpfpRCy4bXu+L1bKGmQWkbNu5U+rRbMPWfkUtMPdnRZh00Aw0hm9mrHNNHnIgq7l4mXrk3R5ShS3j4EF2l9HAkhZlHnNP8VgEpz9ZWf1kx10m3Z8orJc7SNLB82HGYN1iG9qro5nnMm/aBaDTdj5gLYT0otFdFt2l+qwMvLaFQ2S1ixFwI60Gh/Szr74+LJmbHxbIEHbqXmNSv0qmh1ZFaFiIZM5zmlxeSnTUeLPUfOJXzZFDsS0t2mJg0j9IkB3uweW4tYrTQaTJqkMEAE19H0/wo/lj5Fn79/VSehjIw26KpDCjNl+ikapgDbkgnBAMpBMGNpqzBCAmyS3PADVm5s0peguBGU9ZghATZpTnghqzcWSUvQXCjKWswQoLs0hxwQ1burJKXILjRlDUYIUF2aQ64ISt3VslLsKyYMEJ6lBYYZZzYempL0R0BQLZp3WKUcWLrqS1FdwQA6dlaYJRxYuupLUV3BADZpnWLUcaJrae2FN0RAKRna4FRxomtp7YU3RFAluUa0/MzQ416KhUbzrlwm5VOtHBAatt2c28CrhERNpW6TfM/GOuf35LEZLCpMLQfZsAKGVuWB+fVjY/QMc/tYnG4PUcbZsAKyS0kh1yn2tup0/z8sI7XmYVa/7RihsPqioN4tAc0MP3mWQzhTR4QBtqqKbhdcTDSqpjBnOa3uqMOXiEUyA0RBLcrDiJ5QAMTeSGdCoIbIgjurDSncrMHap442dZkjUlarcRKSHa+WrbkmCv7LUKOM81fveoHihhMuW5iS0h2W0P9q4lZBlGWMbJhcHngmCMYMF1whtJUa2LlY5VyOXSQiKxN8+cVUiqT39HlYhLmqpG5gcRz6CCxWZvP+lcTM04h7kQzhsG1rWXBYrFKxISKClp5PYhlwO08eCdzCt14hZ1QUZ3IJsZOcGDeyRwnN9VK7QYas4lxiQRk3skccNtqpXYDjdnEuEQCMu9kDrgbnt8n5ngC3DgQw6GD0oOExuzqhA/SmkbNbPDYhTsxHDrgBoLH7EqDD9KaRs1s8Ezz66ccqEIOHXDBBI/ZlQYfpDWNmp4B8nvl5qZ2HsA9DYjBMIEzLThFyRc2WP5LsEAs/Z7mlxINv+uF2sXKPbDq7ysmzYg07jLMglAQl7Eyo2bQTPLemwgAy2wla0HMTjg2cXdthm6S995EAFhmK1kLYnbCsYm7azN0k7z3JgLAMlvJWhCzE45N3F2boZvkvTcRAJbZStaCmJ1wbOLu2gzdJO+9iQCwzJYmJlax0AXicaiEyZcCJl1e3cGuCcQw6RbQ+uAfqJ7mp6eNVEuDtCffA7X+tGLmWQIRBq9GNuQpxCa0E56X1/0OchjSvV0FGgobpvlTjbkkXr4HRv1pYtooeayOHWD0ZEjQdN3rVq6MCg8I0pi8Nw/7HTsAnwwJmq573ab5rQ68mKBikGAUWWqI+jnXAdhkSNB03euG9huT307lCEb5Ugac40Vxl4PEKLvaPqjDmGUUlCSHneaXo2yHGZUrB6KUPiEnoHYiywzJtQUpFh7WJvYWrH9ZMakTPGR0GpJ9wMVXEHyQydO4yTDN75MQNVOJCkGyD7j4CoIPMnkaNxkWWP3LxMQIVFJ/k7nWnds6WouTCxBS/A5KcSVUB6l1p7eO1uLkAoQ0ze93AqUuQHURax28ZqK0U8epDGZR/zgxUwM6A4y8ijwHddggGZQENUndqUke1mY73MlhzngTP80/uIpyvXC6tlpWFYWapO62nvrbxMQAfMRlTpmJCASJPjeIWJAUJZpII0j0uUHEgqQo0UQaQaLPDSIWJEWJJtIIEn1uELEgKUo0kUaQ6HODiAVJUeKK6Y4B5oAp0ZOdnA4dxIUz5yheT0pgwDtgmuaXCqS6UHEcOtgq6y8T075Zjs9PluGUeQIbpHqAIQu7RsxgDJ48d5OH8NP8Vo/hCtnlFfuAIVHNVjKDMZgLs/60YuZOJ8EDYLsOhn0Y3OyltxYQ3xTiPM4q+dzkYPZJiemtBWyu/O9933u7//O5L3Rf+eqXu8W7726XhZsxPw2/HL0Fnp8mJnd/YPpVA+H7EnFJsW2fogAmSe9EeAoQ3NCAhiEGnkhbS/5zP/wP3eted1oa32GHHdZdcskl3a677iI6P2Rshs/VNOdWOn69023v/5uj2xTAJ2bTnI9800wnbpuu0JAYMnugQlL5m3gPgPznn7+sO/nkU+y7R7kEhz/tiO4bF/9zt/NDdxKLF6Kt5wNg/Dy+NHwMF3LC+Ke/j+lFyjMnTBcprUyQufw+6Gc+/Zlu6YkndGvXrpOIujpqItue+cxndl/72te7HR6ygz86g08ld2mu+UOQKiZypTycLDvYtMXyS2Ltx0CX1BS6KHp1+mF3joA25pI9n3/BGV0N6oCsM86BBkwLIf+FF36pe+lxx3Vr778/dVRv7LSrWgu95FH8vOc/r7vwixd22223XeLYrh4Q64xzkwHTQhh/ubCTDs7x+KdTOb8qcgll5HkSpsFbBXSvG6ZnsUaLsWifGsoOt97kUvhAyn/xRRd3L3rxMd2999yb6qQXPKWmAkXVNxjU/pI/+7Pus5//bLdoZpGvyoWba5qKnHY2uS2E7I0Zru0TA8ciUWynzdW3FdXfrjGrwad3ZnQsA2NU89CWxi6O1CQoLbtxP0Dyf+tb3+qe/7zndXfffY8MWgalxUhjy7IqhbqXnnhSd8H
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" style="height:45px;"></td> 1434<td class="gt_row gt_left" headers="Generative (i.e., Classical Statistics) Primary Goals">Inferring relationships between <strong><em>X</em></strong> and <strong><em>Y</em></strong></td> 1435<td class="gt_row gt_left" headers="Generative (i.e., Classical Statistics) Key Strengths">Interpretability; emphasis on uncertainty around estimates; explanatory power</td> 1436<td class="gt_row gt_left" headers="Generative (i.e., Classical Statistics) Key Limitations">Bounded by statistical assumptions, inattention to variance across samples</td> 1437</tr> 1438<tr class="gt_group_heading_row odd"> 1439<td colspan="4" id="Predictive (i.e., Machine Learning)" class="gt_group_heading" data-quarto-table-cell-role="th" style="border-bottom-width: 1px; border-bottom-style: solid; border-bottom-color: black" scope="colgroup">Predictive (i.e., Machine Learning)</td> 1440</tr> 1441<tr class="gt_row_group_first even"> 1442<td class="gt_row gt_center" headers="Predictive (i.e., Machine Learning) Quantity of Interest"><img data-src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAJoAAADgCAYAAAD7Rn+wAAAMP2lDQ1BJQ0MgUHJvZmlsZQAASImVVwdYU8kWnluSkEBoAQSkhN4EkRpASggt9N5EJSQBQokxEFTs6KKCaxcL2NBVEQUrzYIidhbF3hcLKsq6WLArb1JA133le/N9c+e//5z5z5lz5965A4DacY5IlIeqA5AvLBTHBvvTk1NS6aSnAAcUQAVeQJnDLRAxo6PDASxD7d/Lu+sAkbZX7KVa/+z/r0WDxy/gAoBEQ5zBK+DmQ3wQALyKKxIXAkCU8mZTCkVSDCvQEsMAIV4oxVlyXCXFGXK8V2YTH8uCuB0AJRUOR5wFgOolyNOLuFlQQ7UfYkchTyAEQ
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
1442/vxuic0UzBhjlVjL/+kmn7FPFy11eyaoMGnzHcSPNnsoMQyPeKH+upDyDhqr1Mo8AzCtkckKFNN7kz//o99Fs0vNKsUn+4mD0+3D6RjLPS54tRZWvMBS3zsJLFZ+vEkpR0qNLjXbm+x2a9u5WZUDWGadADRP6qa5wHuXnBzGduC0w/7HQ+KrVhZL+pI9D4orZssWO4oOSbPUWHQ2dhShFi5e25Ioe2jVG+eNkhWnZtuZfd7RYMfGke1gx1qIumqpr4pSaXrnFMPp9MpmEbXb+Gz/yElc9VQeVRDKzV0iq9zhTFCH5zxYRpfVICx+iiwxXd08KNfc4R/l1BsL0bOn5Dzva6PfR3E6CpwAUaZCpjMmycdC1F0UOy6PeeHZEpcm5x5wJG0CuvShyTHV+vL0xdDR4eAydbD1ba8/b5zzI1u5FbTKRjAluhfkvvPBbgx/84MbBkh8uGez2ht3qPqPrQeqablvj7x1w4cCyNm6dmKFAiTNkpe9CIIAzQhEnjmbz1DqI3sKNxSVBmgGGXhfROYn5L7v00sHffPrTMcMBBxww+NGPfjTYZZddkFG3IGjN7r+6xy9DSyPAKL0UJ440frt1RnujtTclDUZIBDWpDrghC3dSyUsQXG9KGoyQIJtUB9yQhTup5CUIrjd1g6uvWjw4+eSTw0SLR49DDz10cMcddwxmzZqVLNoKbSHB96tEXoLgelPSYIQE2aQ64IYs3EklL0FwvSlpMEKCbFIdcEc5+n20tGe6WbEZc+Da664bLDjxxMH69evVHq7WsbAlS9F94AMfGNz277cNdpi5g6+jyERwhMu6IQBIUBtSKf3E2lNasm4IALKRFyal9BNrT3jT257REMWmSehyyN4XMO5xFIWgcsDtud1JNGsTgWniSgdsIuXYOvLfcvO/Dj7+Vx8frFu3LnZJ3o3RHsZqi8rhhx8+uOWWmwfbbbe9dr1xxuh0IkwjJmwafWsZ/+au/+j30eK6YnFpvQnefvvtg6OOPnrw8tq1ZE0XAFnkDeljPvaxwfXXXz+YNn366PfZ7I34tKO5aTbFAE9lE4NpLw6DwX0IAEKzNe1y4jeugZ5W2QzmZOS/8847B38RdqoXXnzRNhdk1lLLOxoICxZ8YnDN4qvzJKCDaFhI5zbFQMGuVTAnY/x1ttoyrvxy6+RDntn0jyGz1dnWq5a9G0KZH1E4ZUvdVn2e4WxTmP+uu3/azdppVpjH+LGFPDzInKouj2b6Rr7aAoZf5Omnn54Gl8cSUThly9Y9fvRO+5t6vYnzL9s7HTQFBg1EHmvIGW3soIgKhzqJTTyDBqY0/z0/v6fbeeedrXi0wFBog27OnNndXnvNz4UnBciFF/Qzzzwzjc2PgQZcQOIZNDCl489Xw8TkLwrNX20+RTEnqSfDObmN8jI7oqwakU2MjWBAvcM5Rq56G9s1Govpl/f+stt1111TEaUdjArpne94R/fQww91zz77bPfhD3/EdrSyGEX/0pe+FDuhqXLCiLJqHWUTYyMYUO9wjpE3avxoNTz2xuW3QusPur7qZDKE/vS16rPLEOCD1GF5TW3wqI8YBg1wg4D77EKDD1KbQrvvvvu63d/4RioyKrRQOO877H3dU797ShuFWC+/vLY78cQFxNfbrO5uir/5zW8mvghkglSX14ge+OojhkED3IByFOaoog2kcrzG7SYmfyi0IoWpBjirw56BDiWKcwYl6bjd5kCOaDwCmVog33Lz8y9btqybM3tO7w519NFHdy+88IL1AvnD+2rdF75wViq2VGiyA6Z/8sx2yT9eEtttzePXDmJ4GB30WnrG8Pm3Ha1c2NwsoyqVz2Tu0gxdJZ+tSQBgqS1nzYjZEfsm5i7N0FXy2Zp0q1at7N7ylrdYcWiR5N1p4cLTuvAeWm4gCIGT9eKLLuqmTZvWjDFtbKy78rvftUZFU7OnUCE0GBnBZxIUMygozdBV8pkbgoUY0Ccmfyw0XGUuJPKEvNy1iMkXvTkA91xbOm7hTmpuTuQKqiGeyTcR+R966OFu/vy9rUDCWzMR41XkP8TnrPHlv+GGG7qZM3ewWLFgQ5GJlCIM77FVk7Clxz8V+WlHK8bvFlN8bEglySZrTuVqfgMpDOnWrgAVhQ0Tl/83v3m822+//eIrRhQW3s6YNm16t2jRotSx8ee/88c/7l63yy6+2OQ2Ggp4xvQZ3ZIf/lsx2IbK6aKbDRM3/kZmNXG6CchPhaaROb5hA+gWGSJUXc7kcRrskIiUpXrYb9gA2GSIUHU5k8dpsENKpCd/92T3znf+Yd69aCebucPM7gc3/YvQyqDeEANqVDkj/r3/e2/35rn6vJcLWHfKmTNndnfc8Z8SmY4cA0bEsqBwsCGScltrE7lZA4K0UAbUw37DBkAmQ4Sqy5k8TtNbZ6N9NGFP5SY+ElqWGXp0NE7Si0Yb8MmVTTm3oNLe1GFc3z399NPdAe86oMNtEruY3OLCn/x0P/nJT0JM8Cl8Nm0w/+rVD3b7779/2tn8i4Qdd9yxu+u/7ooxLKSBFHoz87sOxthIkKQXNMiJz593NBoUd7DoGrsMg1P31ChaB5mYHGSYwvzPhfe/wp/21Le2UGRz58zp7r33Xus4eghpDgLZlxHcTz75ZHfoe9u5dt75td0999yj1CkcP/rm1muS8+dCy9ld/miu5y+xa0dt4cC5GT3J1YQySKlbi9pRW4wcwZoX1nTv/5P3p2cy2mXCbXOfffbpVq5cNeHjX7NmTXfUUUf5wg755Ae8Xv/617vCjp0sB1HqNqTaUVuMnEEgTfX8+0KLHcj94VsHWTOkUSkkg7CgRikn/xALtwUMBmfrucqYD6ztXOscLJrXdy++9EL3kY9+NCx4fttC38YYdAe956Dut0/8dtLyr1v3cnfa6Qv9bTQ9E+6++xu7B5Y+EPvrRjDB49+S86+F5kanS+dNpBHEIm+yRCxICuRNpBEk+gbhupdfDrvK0bar8AP6n330z7vnnn/OxfBpSCPoGoxTOeecr8Q+cH4p/D3m7dGtWLHCovg0pBE08qYCxIKkON5EGkGibxD6Hc3ojWgNU6RHOzkNGrBdIlvyZmMpHWBmcjRMG5P/vPPOsyLT97b0FeCCBQu6tWvXuuzN3m1mfkmAEJdffnk3bcZ01x8pvMMOOyz1A0zqVsMUvdFOToMGLG+25L5QBoLMTOaGaWPyh0ILtzMKQtASwwYpDmBII1eAGYxBnJr8jz/+ePdHf6xvZeB2edZZZ8WRbInxL1mypJNXnujLa3faqQtfcrF5xeyIxKxB9tm4TcatVtk7VetPO1rqUBSNzpnJAPd23NhaB2A4tk5aFOxhuxDZN+60kfjUU091Bx9ySCcfB11w/gVFJM7DOdi+efld6xD2v3/2s+4Nb9it2y28IPj5z/9HBxPTTU1+zmLzOkn5G98ZCNdYuszCdp6PcO3JH2fBhr/mFIJclvpcm7xqSI7W9xlByOFLVDGCYSLyP//884OlD9w/OOBdB5YpnT5Z+TVJjv7ggw/GL7vsueeeWyS/S0pK7mEybu78y2Ukoapy4EqiDgBWfDiiRDchkxMqZCpR8VbxRvnD1YvLOs0fiWq+yJeu8DireWsgdQvM/+j30WzS80qxSa7D0e+jacHzvOTZUlT5CkNx6yy8tOPkq4RSlPToUqOd+X6Hpr27VRmQdcYpUMOEfqornEf53aamSxTmpblblhPKOuONn/9YaHzV6kJJIH7e0iRylgMberZ6i7LoHBuGU3NwISPtGqP8xSzHSY6neNm/Wudfd7RiMPF9cx4R1UwfdNPjlLpF5R7lDxdhnAS9h1QTVM9haXFNnFIyq0tZDZOcv/EjLzpg7R71OJnZKxzVe5wYY3Bv6d/nGuWXj3fLO0/rJQWvsGA5QrueJVZzj1Mbx7ZhRxv9PppMZJyqNF+YH8geM9wblK69KHJgzflSdUSlybnHnAkbQK79lsofno80tetN0XM8vAdO69m6YDs1h2
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" style="height:45px;"></td> 1443<td class="gt_row gt_left" headers="Predictive (i.e., Machine Learning) Primary Goals">Generating accurate predictions of <strong><em>Y</em></strong></td> 1444<td class="gt_row gt_left" headers="Predictive (i.e., Machine Learning) Key Strengths">Predictive power; potential to simplify high dimensional data; relatively unconstrained by statistical assumptions</td> 1445<td class="gt_row gt_left" headers="Predictive (i.e., Machine Learning) Key Limitations">Inattention to explanatory processes, opaque links between <strong><em>X</em></strong> and <strong><em>Y</em></strong></td> 1446</tr> 1447</tbody> 1448</table> 1449 1450</div> 1451</div> 1452</div> 1453<div class="scrollable-note"> 1454<p><span style="font-size: 80%;"><strong>Note</strong>: To be sure, the putative <em>strengths</em> and <em>weaknesses</em> of these modelling âculturesâ have been hotly debated.</span></p> 1455</div> 1456</section> 1457<section id="the-affordances-of-machine-learning" class="slide level2 smaller"> 1458<h2>The Affordances of Machine Learning</h2> 1459<div style="margin-top: 25px;"> 1460 1461</div> 1462<ul> 1463<li class="fragment"><p>Advances in machine learning can provide empirical leverage to social scientists and <em>sharpen</em> social theory in one fell swoop.</p></li> 1464<li class="fragment"><p>Lundberg, Brand and Jeon <span class="citation" data-cites="lundberg_researcher_2022">(<a href="#/references" role="doc-biblioref" onclick="">2022</a>)</span>, for instance, argue that adopting a machine learning framework can help social scientists:</p> 1465<ul> 1466<li class="fragment"><span style="font-size: 85%;">Amplify <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>human coding</strong></span></span></li> 1467<li class="fragment"><span style="font-size: 85%;">Summarize <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>complex data structures</strong></span></span></li> 1468<li class="fragment"><span style="font-size: 85%;">Relax <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>statistical assumptions</strong></span></span></li> 1469<li class="fragment"><span style="font-size: 85%;">Target <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>researcher attention</strong></span></span></li> 1470</ul></li> 1471<li class="fragment"><p>While ML is often associated with <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>induction</strong></span>, van Loon <span class="citation" data-cites="loon_machine_2022">(<a href="#/references" role="doc-biblioref" onclick="">2022</a>)</span> argues that SML algorithms can help us deductively resolve <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>
1471predictability hypotheses</strong></span> as well.</p> 1472<ul> 1473<li class="fragment"><span style="font-size: 85%;"><em>What plays a larger role in shaping human behaviourânature or nurture</em>?</span></li> 1474</ul></li> 1475</ul> 1476</section></section> 1477<section> 1478<section id="key-terms-and-concepts-in-the-sml-setting" class="title-slide slide level1 center" data-background-color="#43464B"> 1479<h1>Key Terms and Concepts in the SML Setting</h1> 1480 1481</section> 1482<section id="bias-variance-tradeoff" class="slide level2 smaller"> 1483<h2>Bias-Variance Tradeoff</h2> 1484<div style="margin-top: 40px;"> 1485 1486</div> 1487<div class="columns"> 1488<div class="column" style="width:62%;"> 1489<p><a href="./media/bias_variance.jpeg" class="lightbox" data-gallery="quarto-lightbox-gallery-2"><img data-src="./media/bias_variance.jpeg"></a></p> 1490<div style="text-align: center;"> 1491<p><span style="font-size: 60%; text-align: center;">Image can be retrieved <a href="https://i.imgur.com/qRvK70r.jpg" class="external" target="_blank">here</a>.</span></p> 1492</div> 1493</div><div class="column" style="width:38%;"> 1494<ul> 1495<li class="fragment"><p><span style="font-size: 75%;"><span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>Bias</strong></span> emerges when we build SML algorithms that fail to sufficiently map the patternsâor pick up the empirical signalâlinking <strong><span class="math inline">X</span></strong> and <strong><span class="math inline">Y</span></strong>. Think: <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>underfitting</strong></span>.</span></p></li> 1496<li class="fragment"><p><span style="font-size: 75%;"><span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>Variance</strong></span> arises when our algorithms not only pick up the signal linking <strong><span class="math inline">X</span></strong> and <strong><span class="math inline">Y</span></strong>, but some of the noise in our data as well. Think: <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>overfitting</strong></span>.</span></p></li> 1497<li class="fragment"><p><span style="font-size: 75%;">When adopting an SML framework, researchers try to <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>strike the optimal balance</em></span> between bias and variance.</span></p></li> 1498</ul> 1499</div></div> 1500</section> 1501<section id="training-validation-and-testing" class="slide level2 smaller"> 1502<h2>Training, Validation and Testing</h2> 1503<div style="margin-top: 30px;"> 1504 1505</div> 1506<div class="fragment"> 1507<ul> 1508<li>In an SML setting, we want to reduce our algorithmâs <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>generalization</em></span> or <span style="text-decoration: underline; text-decoration-color: #5d3c85;"><em>test error</em></span>âi.e., âthe prediction error of a model on new dataâ <span class="citation" data-cites="molina_machine_2019">(<a href="#/references" role="doc-biblioref" onclick="">Molina and Garip 2019</a>)</span>.</li> 1509</ul> 1510</div> 1511<div class="fragment"> 1512<ul> 1513<li>To arrive at an estimate of our modelâs performance, we can (randomly) partition our global sample of observations into disjoint <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>sets</strong></span> or subsamples.</li> 1514</ul> 1515</div> 1516<div class="fragment"> 1517<div class="panel-tabset"> 1518<ul id="tabset-2" class="panel-tabset-tabby"><li><a data-tabby-default="" href="#tabset-2-1">Training</a></li><li><a href="#tabset-2-2">Validation</a></li><li><a href="#tabset-2-3">Testing</a></li></ul> 1519<div class="tab-content"> 1520<div id="tabset-2-1"> 1521<ul> 1522<li class="fragment"><p><span style="font-size: 85%;">We can use a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>training set</strong></span> to fit our algorithmâto find weights (or coefficients), recursively split the feature space to grow decision trees and s
1522o on.</span></p></li> 1523<li class="fragment"><p><span style="font-size: 85%;">Training data should constitute the <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>largest</strong></span> of our three disjoint sets.</span></p></li> 1524</ul> 1525</div> 1526<div id="tabset-2-2"> 1527<ul> 1528<li class="fragment"><p><span style="font-size: 85%;">We can use a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>validation set</strong></span> to find the right estimator out of a series of candidate algorithmsâor select the best-fitting parameterization of a single algorithm.</span></p></li> 1529<li class="fragment"><p><span style="font-size: 85%;">Often, using both training <em>and</em> validation sets can be costly: data sparsity can amplify <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>variance</strong></span> across samples or datasets.</span></p></li> 1530<li class="fragment"><p><span style="font-size: 85%;">Thus, when limited to smaller samples, analysts often <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>combine</strong></span> training and validationâsay, by recycling training data for model tuning and selection.</span></p></li> 1531</ul> 1532</div> 1533<div id="tabset-2-3"> 1534<ul> 1535<li class="fragment"><p><span style="font-size: 85%;">We can use a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>testing set</strong></span> to generate a measure of our modelâs predictive accuracy (e.g., the <code>F1 score</code> for classification problems)âor to derive our generalization error.</span></p></li> 1536<li class="fragment"><p><span style="font-size: 85%;">This subsample is used only <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>once</strong></span> (to report the performance metric); put another way, it <strong>cannot</strong> be used to train, tune or select our algorithm.</span></p></li> 1537</ul> 1538</div> 1539</div> 1540</div> 1541</div> 1542</section> 1543<section id="k-fold-cross-validation" class="slide level2 smaller"> 1544<h2><span class="math inline">k</span>-Fold Cross-Validation</h2> 1545<div class="panel-tabset"> 1546<ul id="tabset-3" class="panel-tabset-tabby"><li><a data-tabby-default="" href="#tabset-3-1">High-Level Overview</a></li><li><a href="#tabset-3-2">An Illustration</a></li></ul> 1547<div class="tab-content"> 1548<div id="tabset-3-1"> 1549<ul> 1550<li class="fragment"><p>Unlike conventional approaches to sample partition, <span style="color: #5d3c85;"><strong><span class="math inline">k</span></strong></span> or <span style="color: #5d3c85;"><strong><span class="math inline">v</span>-fold cross-validation</strong></span> allows us to learn from <em>all</em> our data.</p></li> 1551<li class="fragment"><p><span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><span class="math inline">k</span>-fold cross-validation</strong></span> proceeds as follows:</p> 1552<ul> 1553<li class="fragment"><span style="font-size: 75%;">We randomly divide our overall sample into <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><span class="math inline">k</span> subsets or folds</strong></span>.</span></li> 1554<li class="fragment"><span style="font-size: 75%;">We train our algorithm on <span class="math inline">k - 1</span> folds, holding just <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>one group</strong></span> out for model assessment.</span></li> 1555<li class="fragment"><span style="font-size: 75%;">We <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>repeat</strong></span> this process <span class="math inline">k</span> timesâevery fold is held out once and used to fit the model <span class="math inline">k - 1</span> times.</span></li> 1556<li class="fragment"><span style="font-size: 75%;">We then pool or average the evaluation metrics (e.g., predictive accuracy) for <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>all</strong></span> the held-out runs.</span></li> 1557</ul></li> 1558<li class="fragment"><p><span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>Stratified</strong></span> <span class="math inline">k</span>-fold cross-validation ensures that the distribution of class labels (or for numeric targets, the mean) is relatively constant across folds.</p></li> 1559</ul> 1560</div> 1561<div id="tabset-3-2"> 1562<div style="text-align: center;"> 1563<p><span style="font-size: 80%; text-align: center;">Stylized example of <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><em>five-fold</em></span> cross-validation</span></p> 1564</div> 1565<div class="cell" data-layout-align="center"> 1566<div class="cell-output-display"> 1567<div class="quarto-figure quarto-figure-center"> 1568<figure> 1569<p><a href="index_files/figure-revealjs/unnamed-chunk-9-1.png" class="lightbox" data-gallery="quarto-lightbox-gallery-3"><img data-src="index_files/figure-revealjs/unnamed-chunk-9-1.png" class="quarto-figure quarto-figure-center" style="width:150.0%"></a></p> 1570</figure> 1571</div> 1572</div> 1573</div> 1574</div> 1575</div> 1576</div> 1577<!-- {.absolute top="147" left="150" width="750" height="485"} --> 1578</section> 1579<section id="hyperparameter-optimization" class="slide level2 shrinker"> 1580<h2>Hyperparameter Optimization</h2> 1581<div style="margin-top: 30px;"> 1582 1583</div> 1584<ul> 1585<li class="fragment"><p>In SML settings, we automatically learn the <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>
1585parameters</strong></span> (e.g., coefficients) of our algorithms during estimation.</p></li> 1586<li class="fragment"><p><span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>Hyperparameters</strong></span>, on the other hand, are <em>chosen</em> by the analyst, guide the entire learning or estimation process, and can powerfully shape our algorithmâs predictive performance.</p> 1587<ul> 1588<li class="fragment"><span style="font-size: 75%;">Examples of hyperparameters include the <span style="color: #5d3c85;"><strong><span class="math inline">k</span></strong></span> in nearest neighbours models, the <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><span class="math inline">\alpha</span> penalty term</strong></span> in ridge regressions, or the <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>number of hidden layers</strong></span> in a neural network.</span></li> 1589</ul></li> 1590<li class="fragment"><p>How can analysts settle on the right hyperparameter value(s)?</p> 1591<ul> 1592<li class="fragment"><span style="font-size: 75%;">Test different values via trial and error.</span></li> 1593<li class="fragment"><span style="font-size: 90%;">Use automated procedures like <a href="https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html" class="external" target="_blank"><span style="font-size: 90%;"><code>GridSearchCV</code></span></a> from <span style="font-size: 90%;"><code>scikit-learn</code></span>.</span></li> 1594</ul></li> 1595</ul> 1596</section></section> 1597<section> 1598<section id="k-nearest-neighbours" class="title-slide slide level1 center" data-background-color="#43464B"> 1599<h1><span class="math inline">k</span>-Nearest Neighbours</h1> 1600 1601</section> 1602<section id="brief-overview-of-knn" class="slide level2 smaller"> 1603<h2>Brief Overview of KNN</h2> 1604<div style="margin-top: 40px;"> 1605 1606</div> 1607<ul> 1608<li class="fragment"><p><span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><span class="math inline">k</span>-nearest neighbours</strong></span> (KNNs) are simple, non-parametric algorithms that predict values of <span class="math inline">Y</span> based on the distance between rows (or observationsâ inputs).</p></li> 1609<li class="fragment"><p>The estimation of KNNs proceeds as follows:</p> 1610<ul> 1611<li class="fragment"><span style="font-size: 85%;">The analyst defines the <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>distance metric</strong></span> (e.g., Euclidean, Manhattan) they will use to determine how similar any two observations are based on their <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>vector of features</strong></span>.</span></li> 1612<li class="fragment"><span style="font-size: 80%;">The analyst defines a value for the <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>hyperparameter <span class="math inline">k</span></strong></span>âthat is, the number of nearest neighbours to find in the training subsample.</span><br> 1613</li> 1614<li class="fragment"><span style="font-size: 80%;">When fed a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>new data point</strong></span>, KNNs find the <span class="math inline">k</span> nearest neighbours in the training data, and:</span> 1615<ul> 1616<li class="fragment"><span style="font-size: 80%;">Assign the new observation to the class that most of its <span class="math inline">k</span> nearest neighbours belong to (for <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>classification</strong></span> problems.)</span><br> 1617</li> 1618<li class="fragment"><span style="font-size: 80%;">Generate a prediction of <strong><span class="math inline">Y</span></strong> by taking the average of the target variable for the new observationâs <span class="math inline">k</span> nearest neighbours (for <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>regression</strong></span> problems.)</span><br> 1619</li> 1620</ul></li> 1621</ul></li> 1622</ul> 1623</section> 1624<section id="knn-in-python" class="slide level2"> 1625<h2>KNN in Python <svg aria-hidden="true" role="img" viewbox="0 0 448 512" style="height:1em;width:0.88em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:#5d3c85;overflow:visible;position:relative;"><path d="M439.8 200.5c-7.7-30.9-22.3-54.2-53.4-54.2h-40.1v47.4c0 36.8-31.2 67.8-66.8 67.8H172.7c-29.2 0-53.4 25-53.4 54.3v101.8c0 29 25.2 46 53.4 54.3 33.8 9.9 66.3 11.7 106.8 0 26.9-7.8 53.4-23.5 53.4-54.3v-40.7H226.2v-13.6h160.2c31.1 0 42.6-21.7 53.4-54.2 11.2-33.5 10.7-65.7 0-108.6zM286.2 404c11.1 0 20.1 9.1 20.1 20.3 0 11.3-9 20.4-20.1 20.4-11 0-20.1-9.2-20.1-20.4.1-11.3 9.1-20.3 20.1-20.3zM167.8 248.1h106.8c29.7 0 53.4-24.5 53.4-54.3V91.9c0-29-24.4-50.7-53.4-55.6-35.8-5.9-74.7-5.6-106.8.1-45.2 8-53.4 24.7-53.4 55.6v40.7h106.9v13.6h-147c-31.1 0-58.3 18.7-66.8 54.2-9.8 40.7-10.2 66.1 0 108.6 7.6 31.6 25.7 54.2 56.8 54.2H101v-48.8c0-35.3 30.5-66.4 66.8-66.4zm-6.7-142.6c-11.1 0-20.1-9.1-20.1-20.3.1-11.3 9-20.4 20.1-20.4 11 0 20.1 9.2 20.1 20.4s-9 20.3-20.1 20.3z"></path></svg></h2> 1626<div style="margin-top: 90px; text-align: center;"> 1627<div class="callout callout-tip callout-titled callout-style-default"> 1628<div class="callout-body"> 1629<div class="callout-title"> 1630<div class="callout-icon-container"> 1631<i class="callout-icon"></i> 1632</div> 1633<p><strong>Note</strong></p> 1634</div> 1635<div class="callout-content"> 1636<p>The rest of todayâs session will take place in <a href="https://github.com/sakeefkarim/soci-269-f25/blob/main/code/week%2012/soci_269_w12.ipynb" class="external" target="_blank"><span style="color: #5d3c85;"><i class="fa-brands fa-google" aria-label="google"></i></span> Colab</a>.</p> 1637</div> 1638</div> 1639</div> 1640</div> 1641<div class="invisible"> 1642<p><span class="citation" data-cites="mckinney_python_2022">(<a href="#/references" role="doc-biblioref" onclick="">McKinney 2022</a>)</span></p> 1643</div> 1644</section></section> 1645<section> 1646<section id="an-open-work-session-november-19th" class="title-slide slide level1 center" data-background-color="#43464B"> 1647<h1>An Open Work Sessionâ<br>November 19<span style="font-size: 70%;"><sup>th</sup></span></h1> 1648 1649</section> 1650<section id="reminders-4" class="slide level2"> 1651<h2>Reminders</h2> 1652<div class="centre-text"> 1653<p>You have a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>bit</em></strong></span> more time with<br>the second coding assignment.</p> 1654</div> 1655</section> 1656<section id="reminders-5" class="slide level2"> 1657<h2>Reminders</h2> 1658<div style="margin-top: 90px; text-align: center;"> 1659<div class="callout callout-warning callout-titled callout-style-default"> 1660<div class="callout-body"> 1661<div class="callout-title"> 1662<div class="callout-icon-container"> 1663<i class="callout-icon"></i> 1664</div> 1665<p><strong>Second Coding Assignment <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>Soft</em></strong></span> Deadline</strong></p> 1666</div> 1667<div class="callout-content"> 1668<p>Your second coding assignment is due by <strong>8:00 PM</strong>
1668 on Friday, November 21<span style="font-size:70%"><sup>st</sup></span>.<br>This is a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>soft</em></strong></span> deadline.</p> 1669</div> 1670</div> 1671</div> 1672</div> 1673</section> 1674<section id="reminders-6" class="slide level2"> 1675<h2>Reminders</h2> 1676<div style="margin-top: 90px; text-align: center;"> 1677<div class="callout callout-important callout-titled callout-style-default"> 1678<div class="callout-body"> 1679<div class="callout-title"> 1680<div class="callout-icon-container"> 1681<i class="callout-icon"></i> 1682</div> 1683<p><strong>Second Coding Assignment <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>Hard</em></strong></span> Deadline</strong></p> 1684</div> 1685<div class="callout-content"> 1686<p>All coding assignments (in <code>Python</code> <svg aria-hidden="true" role="img" viewbox="0 0 448 512" style="height:1em;width:0.88em;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:#5d3c85;overflow:visible;position:relative;"><path d="M439.8 200.5c-7.7-30.9-22.3-54.2-53.4-54.2h-40.1v47.4c0 36.8-31.2 67.8-66.8 67.8H172.7c-29.2 0-53.4 25-53.4 54.3v101.8c0 29 25.2 46 53.4 54.3 33.8 9.9 66.3 11.7 106.8 0 26.9-7.8 53.4-23.5 53.4-54.3v-40.7H226.2v-13.6h160.2c31.1 0 42.6-21.7 53.4-54.2 11.2-33.5 10.7-65.7 0-108.6zM286.2 404c11.1 0 20.1 9.1 20.1 20.3 0 11.3-9 20.4-20.1 20.4-11 0-20.1-9.2-20.1-20.4.1-11.3 9.1-20.3 20.1-20.3zM167.8 248.1h106.8c29.7 0 53.4-24.5 53.4-54.3V91.9c0-29-24.4-50.7-53.4-55.6-35.8-5.9-74.7-5.6-106.8.1-45.2 8-53.4 24.7-53.4 55.6v40.7h106.9v13.6h-147c-31.1 0-58.3 18.7-66.8 54.2-9.8 40.7-10.2 66.1 0 108.6 7.6 31.6 25.7 54.2 56.8 54.2H101v-48.8c0-35.3 30.5-66.4 66.8-66.4zm-6.7-142.6c-11.1 0-20.1-9.1-20.1-20.3.1-11.3 9-20.4 20.1-20.4 11 0 20.1 9.2 20.1 20.4s-9 20.3-20.1 20.3z"></path></svg>) <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>must</em></strong></span> be submitted by <strong>8:00 PM</strong> on Wednesday, December 3<span style="font-size:70%"><sup>rd</sup></span>.</p> 1687</div> 1688</div> 1689</div> 1690</div> 1691</section> 1692<section id="reminders-7" class="slide level2"> 1693<h2>Reminders</h2> 1694<div style="text-align: center; margin-top: 25px;"> 1695<p>Once again, assignment instructions are available online.</p> 1696<p align="center"> 1697<iframe class="iframe" src="https://soci-269-a2.netlify.app/" height="450" width="1040" style="border: 1px solid #2e3846;"> 1698</iframe> 1699</p> 1700<div style="text-align: center; font-size: 75%; margin-top:-40px"> 1701<p><a href="https://soci-269-a1.netlify.app/" data-preview-link="true">Full Page</a></p> 1702</div> 1703</div> 1704</section> 1705<section id="reminders-8" class="slide level2"> 1706<h2>Reminders</h2> 1707<h4 id="final-presentation-and-term-paper-1"><span style="color:#a9a9a9;">Final Presentation and Term Paper</span></h4> 1708<div style="text-align: center; font-size: 87%; margin-top: 20px; margin-bottom: 20px;"> 1709<p>Guidelines for your <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>final presentation</em></strong></span> and <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>term paper</em></strong></span> are online.</p> 1710<p align="center"> 1711<iframe class="iframe" src="https://soci-269-final.netlify.app/" height="450" width="1040" style="border: 1px solid #2e3846;"> 1712</iframe> 1713</p> 1714<div style="text-align: center; font-size: 60%; margin-top:-40px"> 1715<p><a href="https://soci-269-final.netlify.app/" data-preview-link="true">Full Page</a></p> 1716</div> 1717</div> 1718</section> 1719<section id="reminders-9" class="slide level2"> 1720<h2>Reminders</h2> 1721<div class="centre-text"> 1722<p>You can find out when youâre presenting<br>via <a href="https://moodle.amherst.edu/mod/resource/view.php?id=982310">Moodle</a>.</p> 1723</div> 1724</section> 1725<section id="an-update-1" class="slide level2"> 1726<h2>An Update</h2> 1727<div class="centre-text"> 1728<p>A quick note on <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>grades</em></strong></span>âand an opportunity for <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>resubmitting</em></strong></span> or<br><span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong><em>re-weighting</em></strong></span> a coding assignment.</p> 1729</div> 1730</section></section> 1731<section> 1732<section id="work-session" class="title-slide slide level1 center" data-background-color="#43464B"> 1733<h1>Work Session</h1> 1734 1735</section> 1736<section id="create-a-pitch" class="slide level2"> 1737<h2>Create a Pitch</h2> 1738<div style="margin-top: 80px; text-align: center;"> 1739<p>Carefully review the guidelines for your term paper: i.e., scan the codebooksâor leverage the interactive tablesâ<a href="https://soci-269-final.netlify.app/" class="external" target="_blank">embedded online</a> to isolate variables of interest. How can you connect these items to the substantive material covered in <a href="https://soci-269.netlify.app/#modulei" class="external" target="_blank"><span style="font-size: 103%;"><code>Module I</code></span></a> of this class?</p> 1740<div style="margin-top: 40px;"> 1741 1742</div> 1743<div class="fragment"> 1744<p>Towards the end of class, weâll hold a <span style="color: white; background-color: #5d3c85; padding: 1px 5px; border: 0.001px solid white; border-radius: 5px;"><strong>brainstorming</strong></span> session where everyone will <em>pitch</em> their preliminary ideas.</p> 1745</div> 1746</div> 1747</section></section> 1748<section> 1749<section id="enjoy-the-break" class="title-slide slide level1 center" data-background-color="#43464B"> 1750<h1>Enjoy the Break</h1> 1751 1752</section> 1753<section id="references" class="slide level2 smaller scrollable"> 1754<h2>References</h2> 1755<div style="text-align: center; margin-bottom: 15px"> 1756<p><span style="color: #9c9c9c; font-size:95%;"><strong>Note</strong>: Scroll <span style="color: #5d3c85;"><i class="fa-solid fa-circle-chevron-down" aria-label="circle-chevron-down"></i></span> to access the entire bibliography</span></p> 1757</div> 1758 1759<div id="refs" class="references csl-bib-body hanging-indent" role="list"> 1760<div id="ref-breiman2001" class="csl-entry" role="listitem"> 1761Breiman, Leo. 2001. <span>âStatistical Modeling: The Two Cultures (with Comments and a Rejoinder by the Author).â</span> <em>Statistical Science</em> 16(3). doi: <a href="https://doi.org/10.1214/ss/1009213726">10.1214/ss/1009213726</a>. 1762</div> 1763<div id="ref-donoho2017" class="csl-entry" role="listitem">
1764Donoho, David. 2017. <span>â50 Years of Data Science.â</span> <em>Journal of Computational and Graphical Statistics</em> 26(4):745â66. doi: <a href="https://doi.org/10.1080/10618600.2017.1384734">10.1080/10618600.2017.1384734</a>. 1765</div> 1766<div id="ref-grimmer2021" class="csl-entry" role="listitem"> 1767Grimmer, Justin, Margaret E. Roberts, and Brandon M. Stewart. 2021. <span>âMachine Learning for Social Science: An Agnostic Approach.â</span> <em>Annual Review of Political Science</em> 24(1):395â419. doi: <a href="https://doi.org/10.1146/annurev-polisci-053119-015921">10.1146/annurev-polisci-053119-015921</a>. 1768</div> 1769<div id="ref-lundberg_researcher_2022" class="csl-entry" role="listitem"> 1770Lundberg, Ian, Jennie E. Brand, and Nanum Jeon. 2022. <span>âResearcher Reasoning Meets Computational Capacity: <span>Machine</span> Learning for Social Science.â</span> <em>Social Science Research</em> 108:102807. doi: <a href="https://doi.org/10.1016/j.ssresearch.2022.102807">10.1016/j.ssresearch.2022.102807</a>. 1771</div> 1772<div id="ref-mckinney_python_2022" class="csl-entry" role="listitem"> 1773McKinney, Wes. 2022. <em><a href="https://wesmckinney.com/book/">Python for Data Analysis: Data Wrangling with Pandas, <span>NumPy</span>, and <span>Jupyter</span></a></em>. 3rd Edition. Sebastopol, CA: OâReilly. 1774</div> 1775<div id="ref-molina_machine_2019" class="csl-entry" role="listitem"> 1776Molina, Mario, and Filiz Garip. 2019. <span>âMachine <span>Learning</span> for <span>Sociology</span>.â</span> <em>Annual Review of Sociology</em> 45(Volume 45, 2019):27â45. doi: <a href="https://doi.org/10.1146/annurev-soc-073117-041106">10.1146/annurev-soc-073117-041106</a>. 1777</div> 1778<div id="ref-loon_machine_2022" class="csl-entry" role="listitem"> 1779van Loon, Austin. 2022. <span>â<a href="https://doi.org/10.31235/osf.io/s7uer">Machine Learning and Deductive Social Science: An Introduction to Predictability Hypotheses</a>.â</span> 1780</div> 1781</div> 1782</section></section> 1783 </div> 1784 <div class="quarto-auto-generated-content" style="display: none;"> 1785<p><img src="media/amherst seal.png" class="slide-logo"></p> 1786<div class="footer footer-default"> 1787 1788</div> 1789</div></div> 1790 1791
1791<script>window.backupDefine = window.define; window.define = undefined;</script>
1791 1792
1792<script src="index_files/libs/revealjs/dist/reveal.js"></script>
1792 1793 <!-- reveal.js plugins --> 1794
1794<script src="index_files/libs/revealjs/plugin/quarto-line-highlight/line-highlight.js"></script>
1794 1795
1795<script src="index_files/libs/revealjs/plugin/pdf-export/pdfexport.js"></script>
1795 1796
1796<script src="index_files/libs/revealjs/plugin/reveal-menu/menu.js"></script>
1796 1797
1797<script src="index_files/libs/revealjs/plugin/reveal-menu/quarto-menu.js"></script>
1797 1798
1798<script src="index_files/libs/revealjs/plugin/reveal-chalkboard/plugin.js"></script>
1798 1799
1799<script src="index_files/libs/revealjs/plugin/quarto-support/support.js"></script>
1799 1800 1801 1802
1802<script src="index_files/libs/revealjs/plugin/notes/notes.js"></script>
1802 1803
1803<script src="index_files/libs/revealjs/plugin/search/search.js"></script>
1803 1804
1804<script src="index_files/libs/revealjs/plugin/zoom/zoom.js"></script>
1804 1805
1805<script>window.define = window.backupDefine; window.backupDefine = undefined;</script>
1805 1806 1807
1807<script> 1808 1809 // Full list of configuration options available at: 1810 // https://revealjs.com/config/ 1811 Reveal.initialize({ 1812'controlsAuto': true, 1813'previewLinksAuto': true, 1814'pdfSeparateFragments': false, 1815'autoAnimateEasing': "ease", 1816'autoAnimateDuration': 1, 1817'autoAnimateUnmatched': true, 1818'jumpToSlide': true, 1819'menu': {"side":"left","useTextContentForMissingTitles":true,"markers":false,"loadIcons":false,"custom":[{"title":"Tools","icon":"<i class=\"fas fa-gear\"></i>","content":"<ul class=\"slide-menu-items\">\n<li class=\"slide-tool-item active\" data-item=\"0\"><a href=\"#\" onclick=\"RevealMenuToolHandlers.fullscreen(event)\"><kbd>f</kbd> Fullscreen</a></li>\n<li class=\"slide-tool-item\" data-item=\"1\"><a href=\"#\" onclick=\"RevealMenuToolHandlers.speakerMode(event)\"><kbd>s</kbd> Speaker View</a></li>\n<li class=\"slide-tool-item\" data-item=\"2\"><a href=\"#\" onclick=\"RevealMenuToolHandlers.overview(event)\"><kbd>o</kbd> Slide Overview</a></li>\n<li class=\"slide-tool-item\" data-item=\"3\"><a href=\"#\" onclick=\"RevealMenuToolHandlers.togglePdfExport(event)\"><kbd>e</kbd> PDF Export Mode</a></li>\n<li class=\"slide-tool-item\" data-item=\"4\"><a href=\"#\" onclick=\"RevealMenuToolHandlers.toggleScrollView(event)\"><kbd>r</kbd> Scroll View Mode</a></li>\n<li class=\"slide-tool-item\" data-item=\"5\"><a href=\"#\" onclick=\"RevealMenuToolHandlers.toggleChalkboard(event)\"><kbd>b</kbd> Toggle Chalkboard</a></li>\n<li class=\"slide-tool-item\" data-item=\"6\"><a href=\"#\" onclick=\"RevealMenuToolHandlers.toggleNotesCanvas(event)\"><kbd>c</kbd> Toggle Notes Canvas</a></li>\n<li class=\"slide-tool-item\" data-item=\"7\"><a href=\"#\" onclick=\"RevealMenuToolHandlers.downloadDrawings(event)\"><kbd>d</kbd> Download Drawings</a></li>\n<li class=\"slide-tool-item\" data-item=\"8\"><a href=\"#\" onclick=\"RevealMenuToolHandlers.keyboardHelp(event)\"><kbd>?</kbd> Keyboard Help</a></li>\n</ul>"}],"openButton":true,"numbers":true,"width":"wide"}, 1820'chalkboard': {"buttons":true}, 1821'smaller': false, 1822 1823 // Display controls in the bottom right corner 1824 controls: false, 1825 1826 // Help the user learn the controls by providing hints, for example by 1827 // bouncing the down arrow when they first encounter a vertical slide 1828 controlsTutorial: false, 1829 1830 // Determines where controls appear, "edges" or "bottom-right" 1831 controlsLayout: 'edges', 1832 1833 // Visibility rule for backwards navigation arrows; "faded", "hidden" 1834 // or "visible" 1835 controlsBackArrows: 'faded', 1836 1837 // Display a presentation progress bar 1838 progress: true, 1839 1840 // Display the page number of the current slide 1841 slideNumber: false, 1842 1843 // 'all', 'print', or 'speaker' 1844 showSlideNumber: 'all', 1845 1846 // Add the current slide number to the URL hash so that reloading the 1847 // page/copying the URL will return you to the same slide 1848 hash: true, 1849 1850 // Start with 1 for the hash rather than 0 1851 hashOneBasedIndex: false, 1852 1853 // Flags if we should monitor the hash and change slides accordingly 1854 respondToHashChanges: true, 1855 1856 // Push each slide change to the browser history 1857 history: true, 1858 1859 // Enable keyboard shortcuts for navigation 1860 keyboard: true, 1861 1862 // Enable the slide overview mode 1863 overview: true, 1864 1865 // Disables the default reveal.js slide layout (scaling and centering) 1866 // so that you can use custom CSS layout 1867 disableLayout: false, 1868 1869 // Vertical centering of slides 1870 center: false, 1871 1872 // Enables touch navigation on devices with touch input 1873 touch: true, 1874 1875 // Loop the presentation 1876 loop: false, 1877 1878 // Change the presentation direction to be RTL 1879 rtl: false, 1880 1881 // see https://revealjs.com/vertical-slides/#navigation-mode 1882 navigationMode: 'linear', 1883 1884 // Randomizes the order of slides each time the presentation loads 1885 shuffle: false, 1886 1887 // Turns fragments on and off globally 1888 fragments: true, 1889 1890 // Flags whether to include the current fragment in the URL, 1891 // so that reloading brings you to the same fragment position 1892 fragmentInURL: false, 1893 1894 // Flags if the presentation is running in an embedded mode, 1895 // i.e. contained within a limited portion of the screen 1896 embedded: false, 1897 1898 // Flags if we should show a help overlay when the questionmark 1899 // key is pressed 1900 help: true, 1901 1902 // Flags if it should be possible to pause the presentation (blackout) 1903 pause: true, 1904 1905 // Flags if speaker notes should be visible to all viewers 1906 showNotes: false, 1907 1908 // Global override for autoplaying embedded media (null/true/false) 1909 autoPlayMedia: null, 1910 1911 // Global override for preloading lazy-loaded iframes (null/true/false) 1912 preloadIframes: null, 1913 1914 // Number of milliseconds between automatically proceeding to the
1915 // next slide, disabled when set to 0, this value can be overwritten 1916 // by using a data-autoslide attribute on your slides 1917 autoSlide: 0, 1918 1919 // Stop auto-sliding after user input 1920 autoSlideStoppable: true, 1921 1922 // Use this method for navigation when auto-sliding 1923 autoSlideMethod: null, 1924 1925 // Specify the average time in seconds that you think you will spend 1926 // presenting each slide. This is used to show a pacing timer in the 1927 // speaker view 1928 defaultTiming: null, 1929 1930 // Enable slide navigation via mouse wheel 1931 mouseWheel: false, 1932 1933 // The display mode that will be used to show slides 1934 display: 'block', 1935 1936 // Hide cursor if inactive 1937 hideInactiveCursor: true, 1938 1939 // Time before the cursor is hidden (in ms) 1940 hideCursorTime: 5000, 1941 1942 // Opens links in an iframe preview overlay 1943 previewLinks: false, 1944 1945 // Transition style (none/fade/slide/convex/concave/zoom) 1946 transition: 'none', 1947 1948 // Transition speed (default/fast/slow) 1949 transitionSpeed: 'default', 1950 1951 // Transition style for full page slide backgrounds 1952 // (none/fade/slide/convex/concave/zoom) 1953 backgroundTransition: 'none', 1954 1955 // Number of slides away from the current that are visible 1956 viewDistance: 3, 1957 1958 // Number of slides away from the current that are visible on mobile 1959 // devices. It is advisable to set this to a lower number than 1960 // viewDistance in order to save resources. 1961 mobileViewDistance: 2, 1962 1963 // The "normal" size of the presentation, aspect ratio will be preserved 1964 // when the presentation is scaled to fit different resolutions. Can be 1965 // specified using percentage units. 1966 width: 1050, 1967 1968 height: 700, 1969 1970 // Factor of the display size that should remain empty around the content 1971 margin: 0.1, 1972 1973 // reveal.js plugins 1974 plugins: [QuartoLineHighlight, PdfExport, RevealMenu, RevealChalkboard, QuartoSupport, 1975 1976 RevealNotes, 1977 RevealSearch, 1978 RevealZoom 1979 ] 1980 }); 1981 </script>
1981 1982 1983
1983<script> 1984 // htmlwidgets need to know to resize themselves when slides are shown/hidden. 1985 // Fire the "slideenter" event (handled by htmlwidgets.js) when the current 1986 // slide changes (different for each slide format). 1987 (function () { 1988 // dispatch for htmlwidgets 1989 function fireSlideEnter() { 1990 const event = window.document.createEvent("Event"); 1991 event.initEvent("slideenter", true, true); 1992 window.document.dispatchEvent(event); 1993 } 1994 1995 function fireSlideChanged(previousSlide, currentSlide) { 1996 fireSlideEnter(); 1997 1998 // dispatch for shiny 1999 if (window.jQuery) { 2000 if (previousSlide) { 2001 window.jQuery(previousSlide).trigger("hidden"); 2002 } 2003 if (currentSlide) { 2004 window.jQuery(currentSlide).trigger("shown"); 2005 } 2006 } 2007 } 2008 2009 // hookup for slidy 2010 if (window.w3c_slidy) { 2011 window.w3c_slidy.add_observer(function (slide_num) { 2012 // slide_num starts at position 1 2013 fireSlideChanged(null, w3c_slidy.slides[slide_num - 1]); 2014 }); 2015 } 2016 2017 })(); 2018 </script>
2018 2019 2020
2020<script id="quarto-html-after-body" type="application/javascript"> 2021 window.document.addEventListener("DOMContentLoaded", function (event) { 2022 const tabsets = window.document.querySelectorAll(".panel-tabset-tabby") 2023 tabsets.forEach(function(tabset) { 2024 const tabby = new Tabby('#' + tabset.id); 2025 }); 2026 const isCodeAnnotation = (el) => { 2027 for (const clz of el.classList) { 2028 if (clz.startsWith('code-annotation-')) { 2029 return true; 2030 } 2031 } 2032 return false; 2033 } 2034 const onCopySuccess = function(e) { 2035 // button target 2036 const button = e.trigger; 2037 // don't keep focus 2038 button.blur(); 2039 // flash "checked" 2040 button.classList.add('code-copy-button-checked'); 2041 var currentTitle = button.getAttribute("title"); 2042 button.setAttribute("title", "Copied!"); 2043 let tooltip; 2044 if (window.bootstrap) { 2045 button.setAttribute("data-bs-toggle", "tooltip"); 2046 button.setAttribute("data-bs-placement", "left"); 2047 button.setAttribute("data-bs-title", "Copied!"); 2048 tooltip = new bootstrap.Tooltip(button, 2049 { trigger: "manual", 2050 customClass: "code-copy-button-tooltip", 2051 offset: [0, -8]}); 2052 tooltip.show(); 2053 } 2054 setTimeout(function() { 2055 if (tooltip) { 2056 tooltip.hide(); 2057 button.removeAttribute("data-bs-title"); 2058 button.removeAttribute("data-bs-toggle"); 2059 button.removeAttribute("data-bs-placement"); 2060 } 2061 button.setAttribute("title", currentTitle); 2062 button.classList.remove('code-copy-button-checked'); 2063 }, 1000); 2064 // clear code selection 2065 e.clearSelection(); 2066 } 2067 const getTextToCopy = function(trigger) { 2068 const codeEl = trigger.previousElementSibling.cloneNode(true); 2069 for (const childEl of codeEl.children) { 2070 if (isCodeAnnotation(childEl)) { 2071 childEl.remove(); 2072 } 2073 } 2074 return codeEl.innerText; 2075 } 2076 const clipboard = new window.ClipboardJS('.code-copy-button:not([data-in-quarto-modal])', { 2077 text: getTextToCopy 2078 }); 2079 clipboard.on('success', onCopySuccess); 2080 if (window.document.getElementById('quarto-embedded-source-code-modal')) { 2081 const clipboardModal = new window.ClipboardJS('.code-copy-button[data-in-quarto-modal]', { 2082 text: getTextToCopy, 2083 container: window.document.getElementById('quarto-embedded-source-code-modal') 2084 }); 2085 clipboardModal.on('success', onCopySuccess); 2086 } 2087 var localhostRegex = new RegExp(/^(?:http|https):\/\/localhost\:?[0-9]*\//); 2088 var mailtoRegex = new RegExp(/^mailto:/); 2089 var filterRegex = new RegExp('/' + window.location.host + '/'); 2090 var isInternal = (href) => { 2091 return filterRegex.test(href) || localhostRegex.test(href) || mailtoRegex.test(href); 2092 } 2093 // Inspect non-navigation links and adorn them if external 2094 var links = window.document.querySelectorAll('a[href]:not(.nav-link):not(.navbar-brand):not(.toc-action):not(.sidebar-link):not(.sidebar-item-toggle):not(.pagination-link):not(.no-external):not([aria-hidden]):not(.dropdown-item):not(.quarto-navigation-tool):not(.about-link)'); 2095 for (var i=0; i<links.length; i++) { 2096 const link = links[i]; 2097 if (!isInternal(link.href)) { 2098 // undo the damage that might have been done by quarto-nav.js in the case of 2099 // links that we want to consider external 2100 if (link.dataset.originalHref !== undefined) { 2101 link.href = link.dataset.originalHref; 2102 } 2103 // default icon 2104 link.classList.add("external"); 2105 } 2106 } 2107 function tippyHover(el, contentFn, onTriggerFn, onUntriggerFn) { 2108 const config = { 2109 allowHTML: true, 2110 maxWidth: 500, 2111 delay: 100,
2112 arrow: false, 2113 appendTo: function(el) { 2114 return el.closest('section.slide') || el.parentElement; 2115 }, 2116 interactive: true, 2117 interactiveBorder: 10, 2118 theme: 'light-border', 2119 placement: 'bottom-start', 2120 }; 2121 if (contentFn) { 2122 config.content = contentFn; 2123 } 2124 if (onTriggerFn) { 2125 config.onTrigger = onTriggerFn; 2126 } 2127 if (onUntriggerFn) { 2128 config.onUntrigger = onUntriggerFn; 2129 } 2130 config['offset'] = [0,0]; 2131 config['maxWidth'] = 700; 2132 window.tippy(el, config); 2133 } 2134 const noterefs = window.document.querySelectorAll('a[role="doc-noteref"]'); 2135 for (var i=0; i<noterefs.length; i++) { 2136 const ref = noterefs[i]; 2137 tippyHover(ref, function() { 2138 // use id or data attribute instead here 2139 let href = ref.getAttribute('data-footnote-href') || ref.getAttribute('href'); 2140 try { href = new URL(href).hash; } catch {} 2141 const id = href.replace(/^#\/?/, ""); 2142 const note = window.document.getElementById(id); 2143 if (note) { 2144 return note.innerHTML; 2145 } else { 2146 return ""; 2147 } 2148 }); 2149 } 2150 let selectedAnnoteEl; 2151 const selectorForAnnotation = ( cell, annotation) => { 2152 let cellAttr = 'data-code-cell="' + cell + '"'; 2153 let lineAttr = 'data-code-annotation="' + annotation + '"'; 2154 const selector = 'span[' + cellAttr + '][' + lineAttr + ']'; 2155 return selector; 2156 } 2157 const selectCodeLines = (annoteEl) => { 2158 const doc = window.document; 2159 const targetCell = annoteEl.getAttribute("data-target-cell"); 2160 const targetAnnotation = annoteEl.getAttribute("data-target-annotation"); 2161 const annoteSpan = window.document.querySelector(selectorForAnnotation(targetCell, targetAnnotation)); 2162 const lines = annoteSpan.getAttribute("data-code-lines").split(","); 2163 const lineIds = lines.map((line) => { 2164 return targetCell + "-" + line; 2165 }) 2166 let top = null; 2167 let height = null; 2168 let parent = null; 2169 if (lineIds.length > 0) { 2170 //compute the position of the single el (top and bottom and make a div) 2171 const el = window.document.getElementById(lineIds[0]); 2172 top = el.offsetTop; 2173 height = el.offsetHeight; 2174 parent = el.parentElement.parentElement; 2175 if (lineIds.length > 1) { 2176 const lastEl = window.document.getElementById(lineIds[lineIds.length - 1]); 2177 const bottom = lastEl.offsetTop + lastEl.offsetHeight; 2178 height = bottom - top; 2179 } 2180 if (top !== null && height !== null && parent !== null) { 2181 // cook up a div (if necessary) and position it 2182 let div = window.document.getElementById("code-annotation-line-highlight"); 2183 if (div === null) { 2184 div = window.document.createElement("div"); 2185 div.setAttribute("id", "code-annotation-line-highlight"); 2186 div.style.position = 'absolute'; 2187 parent.appendChild(div); 2188 } 2189 div.style.top = top - 2 + "px"; 2190 div.style.height = height + 4 + "px"; 2191 div.style.left = 0; 2192 let gutterDiv = window.document.getElementById("code-annotation-line-highlight-gutter"); 2193 if (gutterDiv === null) { 2194 gutterDiv = window.document.createElement("div"); 2195 gutterDiv.setAttribute("id", "code-annotation-line-highlight-gutter"); 2196 gutterDiv.style.position = 'absolute'; 2197 const codeCell = window.document.getElementById(targetCell); 2198 const gutter = codeCell.querySelector('.code-annotation-gutter'); 2199 gutter.appendChild(gutterDiv); 2200 } 2201 gutterDiv.style.top = top - 2 + "px"; 2202 gutterDiv.style.height = height + 4 + "px"; 2203 } 2204 selectedAnnoteEl = annoteEl; 2205 } 2206 }; 2207 const unselectCodeLines = () => { 2208 const elementsIds = ["code-annotation-line-highlight", "code-annotation-line-highlight-gutter"]; 2209 elementsIds.forEach((elId) => { 2210 const div = window.document.getElementById(elId); 2211 if (div) { 2212 div.remove(); 2213 } 2214 }); 2215 selectedAnnoteEl = undefined; 2216 }; 2217 // Handle positioning of the toggle 2218 window.addEventListener( 2219 "resize", 2220 throttle(() => { 2221 elRect = undefined; 2222 if (selectedAnnoteEl) { 2223 selectCodeLines(selectedAnnoteEl); 2224 } 2225 }, 10) 2226 ); 2227 function throttle(fn, ms) { 2228 let throttle = false;
2229 let timer; 2230 return (...args) => { 2231 if(!throttle) { // first call gets through 2232 fn.apply(this, args); 2233 throttle = true; 2234 } else { // all the others get throttled 2235 if(timer) clearTimeout(timer); // cancel #2 2236 timer = setTimeout(() => { 2237 fn.apply(this, args); 2238 timer = throttle = false; 2239 }, ms); 2240 } 2241 }; 2242 } 2243 const annoteTargets = window.document.querySelectorAll('.code-annotation-anchor'); 2244 for (let i=0; i<annoteTargets.length; i++) { 2245 const annoteTarget = annoteTargets[i]; 2246 const targetCell = annoteTarget.getAttribute("data-target-cell"); 2247 const targetAnnotation = annoteTarget.getAttribute("data-target-annotation"); 2248 const contentFn = () => { 2249 const content = window.document.querySelector(selectorForAnnotation(targetCell, targetAnnotation)); 2250 if (content) { 2251 const tipContent = content.cloneNode(true); 2252 tipContent.classList.add("code-annotation-tip-content"); 2253 return tipContent.outerHTML; 2254 } 2255 } 2256 const config = { 2257 allowHTML: true, 2258 content: contentFn, 2259 onShow: (instance) => { 2260 selectCodeLines(instance.reference); 2261 instance.reference.classList.add('code-annotation-active'); 2262 window.tippy.hideAll(); 2263 }, 2264 onHide: (instance) => { 2265 unselectCodeLines(); 2266 instance.reference.classList.remove('code-annotation-active'); 2267 }, 2268 maxWidth: 300, 2269 delay: [50, 0], 2270 duration: [200, 0], 2271 offset: [5, 10], 2272 arrow: true, 2273 trigger: 'click', 2274 appendTo: function(el) { 2275 return el.parentElement.parentElement.parentElement; 2276 }, 2277 interactive: true, 2278 interactiveBorder: 10, 2279 theme: 'light-border', 2280 placement: 'right', 2281 positionFixed: true, 2282 popperOptions: { 2283 modifiers: [ 2284 { 2285 name: 'flip', 2286 options: { 2287 flipVariations: false, // true by default 2288 allowedAutoPlacements: ['right'], 2289 fallbackPlacements: ['right', 'top', 'top-start', 'top-end', 'bottom', 'bottom-start', 'bottom-end', 'left'], 2290 }, 2291 }, 2292 { 2293 name: 'preventOverflow', 2294 options: { 2295 mainAxis: false, 2296 altAxis: false 2297 } 2298 } 2299 ] 2300 } 2301 }; 2302 window.tippy(annoteTarget, config); 2303 } 2304 const findCites = (el) => { 2305 const parentEl = el.parentElement; 2306 if (parentEl) { 2307 const cites = parentEl.dataset.cites; 2308 if (cites) { 2309 return { 2310 el, 2311 cites: cites.split(' ') 2312 }; 2313 } else { 2314 return findCites(el.parentElement) 2315 } 2316 } else { 2317 return undefined; 2318 } 2319 }; 2320 var bibliorefs = window.document.querySelectorAll('a[role="doc-biblioref"]'); 2321 for (var i=0; i<bibliorefs.length; i++) { 2322 const ref = bibliorefs[i]; 2323 const citeInfo = findCites(ref); 2324 if (citeInfo) { 2325 tippyHover(citeInfo.el, function() { 2326 var popup = window.document.createElement('div'); 2327 citeInfo.cites.forEach(function(cite) { 2328 var citeDiv = window.document.createElement('div'); 2329 citeDiv.classList.add('hanging-indent'); 2330 citeDiv.classList.add('csl-entry'); 2331 var biblioDiv = window.document.getElementById('ref-' + cite); 2332 if (biblioDiv) { 2333 citeDiv.innerHTML = biblioDiv.innerHTML; 2334 } 2335 popup.appendChild(citeDiv); 2336 }); 2337 return popup.innerHTML; 2338 }); 2339 } 2340 } 2341 }); 2342 </script>
2342 2343
2343<script>var lightboxQuarto = GLightbox({"closeEffect":"zoom","descPosition":"bottom","loop":false,"openEffect":"zoom","selector":".lightbox"}); 2344 (function() { 2345 let previousOnload = window.onload; 2346 window.onload = () => { 2347 if (previousOnload) { 2348 previousOnload(); 2349 } 2350 lightboxQuarto.on('slide_before_load', (data) => { 2351 const { slideIndex, slideNode, slideConfig, player, trigger } = data; 2352 const href = trigger.getAttribute('href'); 2353 if (href !== null) { 2354 const imgEl = window.document.querySelector(`a[href="${href}"] img`); 2355 if (imgEl !== null) { 2356 const srcAttr = imgEl.getAttribute("src"); 2357 if (srcAttr && srcAttr.startsWith("data:")) { 2358 slideConfig.href = srcAttr; 2359 } 2360 } 2361 } 2362 }); 2363 2364 lightboxQuarto.on('slide_after_load', (data) => { 2365 const { slideIndex, slideNode, slideConfig, player, trigger } = data; 2366 if (window.Quarto?.typesetMath) { 2367 window.Quarto.typesetMath(slideNode); 2368 } 2369 }); 2370 2371 }; 2372 2373 })(); 2374 </script>
2374 2375 2376 2377</body></html>
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.