1<!doctype html> 2<html lang="en-US"> 3 4 5 6<head> 7 <!-- Global site tag (gtag.js) - Google Analytics -->
vendor: 66 bytes, lines 7-8
7 8 <script async src="https://www.googletagmanager.com/gtag/js?id=
8UA-5541738-27
vendor: 14 bytes, lines 8-9
8"></script> 9
9<script> 10
vendor: 140 bytes, lines 10-14
10window.dataLayer = window.dataLayer || []; 11 function gtag(){dataLayer.push(arguments);} 12 gtag('js', new Date()); 13 14 gtag('config', '
14UA-5541738-27
vendor: 6 bytes, lines 14-15
14'); 15
15</script>
15 16 17 <meta charset="utf-8"> 18 <title>Analyzing online safety through app store reviews from The Washington Post</title> 19 <link href="/font-awesome/css/all.min.css?v=500d1a92f875b1d96d37a3a3f8f0438c" rel="stylesheet"> 20 <!-- <link href="https://cdnjs.cloudflare.com/ajax/libs/bulma/0.7.5/css/bulma.min.css" rel="stylesheet"> 21 <link href="https://fonts.googleapis.com/css?family=Raleway:400,700|Open+Sans:400,700&display=swap" rel="stylesheet"> --> 22 <link rel="stylesheet" href="/css/spectre.min.css?v=5cd401d486f79e82913923fe7d7f47ff"> 23 <link rel="stylesheet" href="/css/spectre-exp.min.css?v=5909d80638a6ae6aa3a455b6f6a6d768"> 24 <link rel="stylesheet" href="/css/spectre-icons.min.css?v=56b1bd38b79450b37939f8adb811d4cd"> 25 26 <link rel="stylesheet" href="/css/tocbot.css?v=e8f0173e7c5216e5359587a88a570b77"> 27 28 <!--
28<script src="https://unpkg.com/lunr/lunr.js"></script>
28 --> 29
29<script src="/js/lunr.js?v=b93de9e9609074d666f5123e866e25ed"></script>
29 30 <!--
30<script src="https://cdnjs.cloudflare.com/ajax/libs/require.js/2.3.6/require.min.js"></script>
30 --> 31
31<script src="/js/require.min.js?v=18823f6a6d208ee1e361bb266ab794d5"></script>
31 32 33 <meta name="viewport" content="width=device-width, initial-scale=1"> 34 <link href="/css/style.css?v=10420579fc48bad7c182d7d5e44a2cbb" rel="stylesheet"> 35 <link href="/css/highlight.css?v=bdd372d828c6988b6071bb877211ebcb" rel="stylesheet"> 36 37 <meta name="description" content="A Natural Language Processing project: After downloading over a hundred thousand reviews of random chat apps, how to find reports of bullying, racism, and unwanted sexual behavior."> 38 <meta content="investigate.ai: Data Science for Journalists" property="og:site_name"> 39 40 <meta name="twitter:creator" content="@dangerscarf"> 41 <meta name="twitter:title" content="Analyzing online safety through app store reviews from The Washington Post"> 42 <meta name="twitter:description" content="A Natural Language Processing project: After downloading over a hundred thousand reviews of random chat apps, how to find reports of bullying, racism, and unwanted sexual behavior."> 43 <meta name="twitter:image" content="https://investigate.ai/images/flame/flame-searching.png"> 44 <meta name="twitter:image:alt" content="A person metaphorically examining data"> 45 <meta name="twitter:card" content="summary" /> 46 47 <meta property="og:type" content="website" /> 48 <meta property="og:url" content="https://investigate.ai/wapo-app-reviews/" /> 49 <meta property="og:title" content="Analyzing online safety through app store reviews from The Washington Post" /> 50 <meta property="og:description" content="A Natural Language Processing project: After downloading over a hundred thousand reviews of random chat apps, how to find reports of bullying, racism, and unwanted sexual behavior." /> 51 <meta property="og:image" content="https://investigate.ai/images/flame/flame-searching.png" /> 52 <meta property="og:image:height" content="912" /> 53 <meta property="og:image:width" content="921" /> 54 <meta property="fb:admins" content="1504164"/> 55 56 <link rel="apple-touch-icon" sizes="180x180" href="/apple-touch-icon.png"> 57 <link rel="icon" type="image/png" sizes="32x32" href="/favicon-32x32.png"> 58 <link rel="icon" type="image/png" sizes="16x16" href="/favicon-16x16.png"> 59 <link rel="manifest" href="/site.webmanifest"> 60 <link rel="mask-icon" href="/safari-pinned-tab.svg" color="#5bbad5"> 61 <meta name="msapplication-TileColor" content="#da532c"> 62 <meta name="theme-color" content="#ffffff"> 63 64</head> 65 66<body class=""> 67 <div> 68 <div class="content-main"> 69 <div class="nav-holder bg-secondary"> 70 <div class='content'> 71 <header class="navbar"> 72 <a href="#" class="sidebar-opener" aria-label="Sidebar opener"><i class="fas fa-bars fa-2x"></i></a> 73 <section class="navbar-section"> 74 <a href="/" class="btn btn-link">Home</a> 75 <a href="/projects/" class="btn btn-link">Projects</a> 76 <a href="/topics/" class="btn btn-link">Topics</a> 77 <a href="/search/" class="btn btn-link">Search</a> 78 <a href="/newsletter/" class="btn btn-link">Newsletter</a> 79 <a href="/about/" class="btn btn-link">About</a> 80 </section> 81 </header> 82 </div> 83</div> 84 85<div class="hero"> 86 <div class="hero-body"> 87 <div class="content"> 88 <h1>Analyzing online safety through app store reviews</h1> 89 <p>After downloading over a hundred thousand reviews of "random chat apps," how to find reports of bullying, racism, and unwanted sexual behavior.</p> 90 91 92 <p> 93 94 <span class="chip">natural language processing</span> 95 96 <span class="chip">text analysis</span> 97 98 <span class="chip">classification</span> 99 100 <span class="chip">reading lots of documents</span> 101 102 </p> 103 104 </div> 105 </div> 106</div> 107<div class="chapter-nav bg-secondary"> 108 <div class="content"> 109 110 111 </div> 112</div> 113<section class="section"> 114 <div class="content"> 115 116 <div class="columns"> 117 <div class="column col-5 col-md-12 readings"> 118 <h3>Readings and links</h3> 119 <ul> 120<li><a href="https://www.washingtonpost.com/technology/2019/11/22/apple-says-its-app-store-is-safe-trusted-place-we-found-reports-unwanted-sexual-behavior-six-apps-some-targeting-minors/">Apple says its App Store is âa safe and trusted place.â We found 1,500 reports of unwanted sexual behavior on six apps, some targeting minors.</a>, from the Washington Post (be sure to watch the video!)</li> 121</ul> 122 </div> 123 <div class="column col-7 col-md-12"> 124 <h3>Summary</h3> 125 <p>The Washington Post downloaded over 130,000 reviews of "random chat apps," investigating cases of bullying, racism, and unwanted sexual behavior. Instead of reading each and every review, they used machine learning to tag likely offenders, resulting in uncovering several thousand potential reviews worth reviewing. In the end they found 1,500 reviews that included "uncomfortable sexual situations." As a result of this research, they could make statements like "at least 19 percent of the reviews on ChatLive mentioned unwanted sexual approaches."</p> 126<p>This project is very similar to the <a href="/nyt-takata-airbags/">Takata airbags project</a>. We'll repeat the same sort of steps:</p> 127<ul> 128<li>Obtain the reviews</li> 129<li>Read a sample of reviews, labeling each as "interesting" or not</li> 130<li>Convert the words in each review to features</li> 131<li>Use the features and labels to train a classifier</li> 132<li>Use the classifier on the unread reviews</li> 133<li>Manually read the ones predicted as interesting</li> 134</ul> 135<p>This project also introduces the concept of <strong>probability</strong> or <strong>decision functions</strong>, where instead of paying attention to the yes/no predicted class we pay attention to the certainty of the prediction.</p> 136<p>If you'd like to increase the performance of the classifier, a good first step would be to flag more reviews.</p> 137 </div> 138 </div> 139 140 141 <h3>Notebooks, Assignments, and Walkthroughs</h3> 142 143 144 145 146 <div class="columns"> 147 <div class="column col-8 col-sm-12"> 148 <h4><a href="/wapo-app-reviews/scrape-app-store-reviews">Scrape and combine app store reviews</a></h4> 149 <p>Using a website devoted to phone app marketing, we'll download over 50,000 reviews for various apps and save them to a CSV.</p> 150 </div> 151 <div class="column col-4 col-sm-12 access-link"> 152 153 154 155 <div class="tile tile-centered"> 156 <div class="tile-icon"> 157 <div class="example-tile-icon"><i class="fa fa-book fa-sm"></i></div> 158 </div> 159 <div class="tile-content"> 160 <div class="tile-title"><a href="/wapo-app-reviews/scrape-app-store-reviews/">Read online</a></div> 161 <small class="tile-subtitle text-gray"> 162 163 Jupyter Notebook 164 </small> 165 </div> 166</div> 167 168 169 170 171 <div class="tile tile-centered"> 172 <div class="tile-icon"> 173 <div class="example-tile-icon"><i class="fa fa-download fa-sm"></i></div> 174 </div> 175 <div class="tile-content"> 176 <div class="tile-title"><a href="/wapo-app-reviews/notebooks/Scrape app store reviews.ipynb">Download notebook</a></div> 177 <small class="tile-subtitle text-gray"> 178 179 Jupyter Notebook 180 </small> 181 </div> 182</div> 183 184 185 186 187 <div class="tile tile-centered"> 188 <div class="tile-icon"> 189 <div class="example-tile-icon"><i class="fa fa-laptop fa-sm"></i></div> 190 </div> 191 <div class="tile-content"> 192 <div class="tile-title"><a href="https://colab.research.google.com/github/littlecolumns/ds4j-notebooks/blob/master/wapo-app-reviews/notebooks/Scrape app store reviews.ipynb">Interactive version</a></div> 193 <small class="tile-subtitle text-gray"> 194 195 Jupyter Notebook 196 </small> 197 </div> 198</div> 199 </div> 200</div> 201 <div class="divider"></div> 202 203 <div class="columns"> 204 <div class="column col-8 col-sm-12"> 205 <h4><a href="/wapo-app-reviews/predict-reviews">Build a classifier to detect reviews about bad behavior</a></h4> 206 <p>Using a small dataset of tagged reviews, can we detect reviews that mention bullying, racism, or unwanted sexual behavior?</p> 207 </div> 208 <div class="column col-4 col-sm-12 access-link"> 209 210 211 212 <div class="tile tile-centered"> 213 <div class="tile-icon"> 214 <div class="example-tile-icon"><i class="fa fa-book fa-sm"></i></div> 215 </div> 216 <div class="tile-content"> 217 <div class="tile-title"><a href="/wapo-app-reviews/predict-reviews/">Read online</a></div> 218 <small class="tile-subtitle text-gray"> 219 220 Jupyter Notebook 221 </small> 222 </div> 223</div> 224 225 226 227 228 <div class="tile tile-centered"> 229 <div class="tile-icon"> 230 <div class="example-tile-icon"><i class="fa fa-download fa-sm"></i></div> 231 </div> 232 <div class="tile-content"> 233 <div class="tile-title"><a href="/wapo-app-reviews/notebooks/Predict reviews.ipynb">Download notebook</a></div> 234 <small class="tile-subtitle text-gray"> 235 236 Jupyter Notebook 237 </small> 238 </div> 239</div> 240 241 242 243 244 <div class="tile tile-centered"> 245 <div class="tile-icon"> 246 <div class="example-tile-icon"><i class="fa fa-laptop fa-sm"></i></div> 247 </div> 248 <div class="tile-content"> 249 <div class="tile-title"><a href="https://colab.research.google.com/github/littlecolumns/ds4j-notebooks/blob/master/wapo-app-reviews/notebooks/Predict reviews.ipynb">Interactive version</a></div> 250 <small class="tile-subtitle text-gray"> 251 252 Jupyter Notebook 253 </small> 254 </div> 255</div> 256 </div> 257</div> 258 <div class="divider"></div> 259 260 261 262 263 264 <div class="columns"> 265 <div class="column col-sm-8"> 266 <h3>Discussion topics</h3> 267 <p>Is it okay to "steal" all of those reviews from Apple?</p> 268<p>What's the difference between using interesting/not interesting flags compared to the chance of a comment being interesting?</p> 269<p>What counts as unwanted sexual behavior or bullying? Find a few examples you feel are borderline.</p> 270<p>In line with the last question, many of these being suggestive/bulling or not are either borderline or a personal decision. In the Takata airbags story, the main drive was to find people to interview. In this case, it's writing sentences like "At least 19 percent of the reviews on ChatLive mentioned unwanted sexual approaches." Is there a difference between the two? Compared to the NYT one, should the Washington Post have taken any additional precautions as a result?</p> 271<p>How can we test whether our classifier does a good job or not? We spent a lot of time testing in the airbags example, but not here. Is there a difference?</p> 272 </div> 273 </div> 274 </div> 275</section> 276 277 <div class="footer bg-secondary"> 278 <div class="content"> 279 <div class="columns"> 280 <div class="column col-8 col-sm-12"> 281 <p><strong>About the site</strong></p> 282 <p>Hi, I'm <a href="https://twitter.com/dangerscarf">Soma</a>, welcome to Data Science for Journalism a.k.a. investigate.ai!</p> 283 <p>There's been a lot of buzz about machine learning and "artificial intelligence" being used in stories over the past few years. It's mostly not that complicated - a little stats, a classifier here or there - but it's hard to know where to start without a little help.</p> 284 <p>If you know a little Python programming, hopefully this site can be that help! <a href="/about">Learn more about this project here.</a></p> 285 <p><strong>Our newsletter</strong></p> 286 <form action="https://littlecolumns.us12.list-manage.com/subscribe/post?u=ecdebf156be0b7e068fac7c25&id=16537e7c90" method="post" role="form" target="_blank" novalidate> 287 <div class="input-group"> 288 <input name="EMAIL" type="email" class="form-input" placeholder="Enter your Email" required=""> 289 <div style="position: absolute; left: -5000px;" aria-hidden="true"><input type="text" name="b_ecdebf156be0b7e068fac7c25_16537e7c90" tabindex="-1" value=""></div> 290 <div class="input-group-append"> 291 <button class="btn btn-primary input-group-btn" type="submit" aria-label="Email signup submit button"> 292 <i class="fas fa-envelope"></i> 293 Sign up 294 </button> 295 </div> 296 </div> 297 </form> 298 299 </div> 300 <div class="column col-4 col-sm-12"> 301 <p><strong>Links</strong></p> 302 <ul> 303 <li><a href="mailto:[email protected]">[email protected]</a></li> 304 <li><a href="https://twitter.com/dangerscarf">@dangerscarf</a></li> 305 <li><a href="/privacy-policy/">Privacy policy</a></li> 306 <li><a href="/newsletter/">Newsletter</a></li> 307 <li>Images via <a href="https://icons8.com/">icons8</a></li> 308 </ul> 309 <p>Thanks to <a href="https://journalism.columbia.edu/">Columbia Journalism School</a>, the <a href="https://knightfoundation.org/">Knight Foundation</a>, and <a href="/about#thankyou">many others</a>.</p> 310 </div> 311 </div> 312 </div> 313</div>
314<script src="/js/scripts.js?v=e9d8cec8fc9221c9dd15f6e0314372b3" type="text/javascript"></script>
314 315 </div> 316 <div class="bg-dark content-sidebar"> 317 <div class="sidebar-holder"> 318 <div class="brand"> 319 <a href="/">investigate.ai</a> 320 <br> 321 data science for everybody 322 </div> 323 324 <div class="accordion-container"> 325 <div class="accordion"> 326 <div class="form-group sidebar-sticky"> 327 <form method="GET" action="/search/" class="search-form"> 328 <div class="input-group"> 329 <input type="text" class="form-input input-sm" name="q" placeholder="Search investigate.ai"> 330 <button class="btn btn-primary input-group-btn btn-sm">Search</button> 331 </div> 332 </form> 333 </div> 334 335 336 <h4 id="textanalysis" class="sidebar-sticky"> 337 <a href="#textanalysis">Text analysis</a> 338 </h4> 339 <ol class="menu menu-nav"> 340 341 <li class="menu-item"> 342 343 <a href="/text-analysis/types-of-text-analysis/"> 344 Types of text analysis 345 </a> 346 <div class="toc"></div> 347 348 349 350 </li> 351 352 <li class="menu-item"> 353 354 355 356 <input id="accordion-countingwords" type="checkbox" name="accordion-checkbox" hidden=""> 357 <label class="accordion-header c-hand" for="accordion-countingwords"> 358 <i class="icon icon-arrow-right mr-1"></i> 359 Counting words 360 </label> 361 <div class="accordion-body"> 362 <ol class="menu menu-nav"> 363 364 <li class="menu-item"> 365 366 <a href="/text-analysis/counting-words-with-pythons-counter/"> 367 Simple word counting 368 </a> 369 <div class="toc"></div> 370 371 372 373 </li> 374 375 <li class="menu-item"> 376 377 <a href="/text-analysis/counting-words-with-scikit-learns-countvectorizer/"> 378 Counting words across many documents 379 </a> 380 <div class="toc"></div> 381 382 383 384 </li> 385 386 <li class="menu-item"> 387 388 <a href="/text-analysis/splitting-words-in-east-asian-languages/"> 389 Segmenting words in East Asian languages 390 </a> 391 <div class="toc"></div> 392 393 394 395 </li> 396 397 <li class="menu-item"> 398 399 <a href="/caixin-museum-word-count/counting-words-in-chinese-museum-names/"> 400 Project: Caixin museums 401 </a> 402 <div class="toc"></div> 403 404 405 406 </li> 407 408 <li class="menu-item"> 409 410 <a href="/text-analysis/how-to-make-scikit-learn-natural-language-processing-work-with-japanese-chinese/"> 411 Using scikit-learn vectorizers with East Asian languages 412 </a> 413 <div class="toc"></div> 414 415 416 417 </li> 418 419 </ol> 420 </div> 421 422 </li> 423 424 <li class="menu-item"> 425 426 427 428 <input id="accordion-advancedwordanalysis" type="checkbox" name="accordion-checkbox" hidden=""> 429 <label class="accordion-header c-hand" for="accordion-advancedwordanalysis"> 430 <i class="icon icon-arrow-right mr-1"></i> 431 Advanced word analysis 432 </label> 433 <div class="accordion-body"> 434 <ol class="menu menu-nav"> 435 436 <li class="menu-item"> 437 438 <a href="/text-analysis/a-simple-explanation-of-tf-idf/"> 439 Upgraded word counts with TF-IDF 440 </a> 441 <div class="toc"></div> 442 443 444 445 </li> 446 447 <li class="menu-item"> 448 449 <a href="/text-analysis/explaining-n-grams-in-natural-language-processing/"> 450 Multi-word phrases and n-grams 451 </a> 452 <div class="toc"></div> 453 454 455 456 </li> 457 458 <li class="menu-item"> 459 460 <a href="/text-analysis/stemming-and-lemmatization/"> 461 Standardizing text with stemming and lemmatization 462 </a> 463 <div class="toc"></div> 464 465 466 467 </li> 468 469 <li class="menu-item"> 470 471 <a href="/text-analysis/using-tf-idf-with-chinese/"> 472 Using TF-IDF with Chinese text 473 </a> 474 <div class="toc"></div> 475 476 477 478 </li> 479 480 </ol> 481 </div> 482 483 </li> 484 485 <li class="menu-item"> 486 487 488 489 <input id="accordion-sentimentanalysis" type="checkbox" name="accordion-checkbox" hidden=""> 490 <label class="accordion-header c-hand" for="accordion-sentimentanalysis"> 491 <i class="icon icon-arrow-right mr-1"></i> 492 Sentiment analysis 493 </label> 494 <div class="accordion-body"> 495 <ol class="menu menu-nav"> 496 497 <li class="menu-item"> 498 499 <a href="/investigating-sentiment-analysis/comparing-sentiment-analysis-tools/"> 500 Comparing sentiment analysis tools 501 </a> 502 <div class="toc"></div> 503 504 505 506 </li> 507 508 <li class="menu-item"> 509 510 <a href="/investigating-sentiment-analysis/designing-your-own-sentiment-analysis-tool/"> 511 Design your own sentiment analyzer 512 </a> 513 <div class="toc"></div> 514 515 516 517 </li> 518 519 <li class="menu-item"> 520 521 <a href="/investigating-sentiment-analysis/more-data-to-train-our-sentiment-analysis-tool/"> 522 Improving your tool 523 </a> 524 <div class="toc"></div> 525 526 527 528 </li> 529 530 <li class="menu-item"> 531 532 <a href="/upshot-trump-emolex/nrc-emotional-lexicon/"> 533 NRC Emotional Lexicon 534 </a> 535 <div class="toc"></div> 536 537 538 539 </li> 540 541 <li class="menu-item"> 542 543 <a href="/upshot-trump-emolex/trump-vs-state-of-the-union-addresses/"> 544 Project: UpShot State of the Union 545 </a> 546 <div class="toc"></div> 547 548 549 550 </li> 551 552 <li class="menu-item"> 553 554 <a href="/nyt-trump-tweets/"> 555 Project: NYT Trump tweets 556 </a> 557 <div class="toc"></div> 558 559 560 561 </li> 562 563 </ol> 564 </div> 565 566 </li> 567 568 <li class="menu-item"> 569 570 571 572 <input id="accordion-documentstotext" type="checkbox" name="accordion-checkbox" hidden=""> 573 <label class="accordion-header c-hand" for="accordion-documentstotext"> 574 <i class="icon icon-arrow-right mr-1"></i> 575 Documents to text 576 </label> 577 <div class="accordion-body"> 578 <ol class="menu menu-nav"> 579 580 <li class="menu-item"> 581 582 <a href="/text-analysis/processing-documents-with-apache-tika/"> 583 Converting documents to text (English) 584 </a> 585 <div class="toc"></div> 586 587 588 589 </li> 590 591 <li class="menu-item"> 592 593 <a href="/text-analysis/processing-documents-with-apache-tika-greek/"> 594 Converting documents to text (non-English) 595 </a> 596 <div class="toc"></div> 597 598 599 600 </li> 601 602 </ol> 603 </div> 604 605 </li> 606 607 <li class="menu-item"> 608 609 610 611 <input id="accordion-conceptspeopleandplaces" type="checkbox" name="accordion-checkbox" hidden=""> 612 <label class="accordion-header c-hand" for="accordion-conceptspeopleandplaces"> 613 <i class="icon icon-arrow-right mr-1"></i> 614 Concepts, people and places 615 </label> 616 <div class="accordion-body"> 617 <ol class="menu menu-nav"> 618 619 <li class="menu-item"> 620 621 <a href="/text-analysis/introduction-to-topic-modeling/"> 622 Extracting topics from documents 623 </a> 624 <div class="toc"></div> 625 626 627 628 </li> 629 630 <li class="menu-item"> 631 632 <a href="/text-analysis/choosing-the-right-number-of-topics-for-a-scikit-learn-topic-model/"> 633 Choosing the right number of topics 634 </a> 635 <div class="toc"></div> 636 637 638 639 </li> 640 641 <li class="menu-item"> 642 643 <a href="/text-analysis/topic-models-with-gensim/"> 644 Topic models with Gensim 645 </a> 646 <div class="toc"></div> 647 648 649 650 </li> 651 652 <li class="menu-item"> 653 654 <a href="/text-analysis/topic-modeling-and-clustering/"> 655 Topic models vs clustering
656 </a> 657 <div class="toc"></div> 658 659 660 661 </li> 662 663 <li class="menu-item"> 664 665 <a href="/text-analysis/named-entity-recognition/"> 666 Entity recognition 667 </a> 668 <div class="toc"></div> 669 670 671 672 </li> 673 674 <li class="menu-item"> 675 676 <a href="/text-analysis/word-embeddings/"> 677 Intro to word embeddings 678 </a> 679 <div class="toc"></div> 680 681 682 683 </li> 684 685 <li class="menu-item"> 686 687 <a href="/text-analysis/document-similarity-using-word-embeddings/"> 688 Conceptual document similarity 689 </a> 690 <div class="toc"></div> 691 692 693 694 </li> 695 696 <li class="menu-item"> 697 698 <a href="/text-analysis/comparing-documents-in-different-languages/"> 699 Comparing documents in different languages 700 </a> 701 <div class="toc"></div> 702 703 704 705 </li> 706 707 </ol> 708 </div> 709 710 </li> 711 712 </ol> 713 714 <h4 id="puttingthingsincategoriesautomatically" class="sidebar-sticky"> 715 <a href="#puttingthingsincategoriesautomatically">Putting things in categories automatically</a> 716 </h4> 717 <ol class="menu menu-nav"> 718 719 <li class="menu-item"> 720 721 <a href="/classification/intro-to-classification/"> 722 Introduction to Classification 723 </a> 724 <div class="toc"></div> 725 726 727 728 </li> 729 730 <li class="menu-item"> 731 732 733 734 <input id="accordion-techniques" type="checkbox" name="accordion-checkbox" hidden=""> 735 <label class="accordion-header c-hand" for="accordion-techniques"> 736 <i class="icon icon-arrow-right mr-1"></i> 737 Techniques 738 </label> 739 <div class="accordion-body"> 740 <ol class="menu menu-nav"> 741 742 <li class="menu-item"> 743 744 <a href="/classification/evaluating-classifiers/"> 745 Evaluating classifiers 746 </a> 747 <div class="toc"></div> 748 749 750 751 </li> 752 753 <li class="menu-item"> 754 755 <a href="/classification/scikit-learn-and-categorical-features/"> 756 Categorical features 757 </a> 758 <div class="toc"></div> 759 760 761 762 </li> 763 764 <li class="menu-item"> 765 766 <a href="/classification/using-classification-algorithms-with-text/"> 767 Classifiers with text 768 </a> 769 <div class="toc"></div> 770 771 772 773 </li> 774 775 <li class="menu-item"> 776 777 <a href="/classification/correcting-for-imbalanced-datasets/"> 778 Correcting for imbalanced datasets 779 </a> 780 <div class="toc"></div> 781 782 783 784 </li> 785 786 </ol> 787 </div> 788 789 </li> 790 791 <li class="menu-item"> 792 793 794 795 <input id="accordion-projects" type="checkbox" name="accordion-checkbox" hidden=""> 796 <label class="accordion-header c-hand" for="accordion-projects"> 797 <i class="icon icon-arrow-right mr-1"></i> 798 Projects 799 </label> 800 <div class="accordion-body"> 801 <ol class="menu menu-nav"> 802 803 <li class="menu-item"> 804 805 <a href="/buzzfeed-spy-planes/buzzfeed-surveillance-planes-random-forests/"> 806 BuzzFeed: Spy planes 807 </a> 808 <div class="toc"></div> 809 810 811 812 </li> 813 814 <li class="menu-item"> 815 816 <a href="/wapo-app-reviews/predict-reviews/"> 817 WaPo chat: App reviews 818 </a> 819 <div class="toc"></div> 820 821 822 823 </li> 824 825 <li class="menu-item"> 826 827 <a href="/nyt-takata-airbags/nyt-takata-completed/"> 828 NYT: Faulty airbag search 829 </a> 830 <div class="toc"></div> 831 832 833 834 </li> 835 836 <li class="menu-item"> 837 838 <a href="/latimes-crime-classification/using-a-classifier-to-find-misclassified-crimes/"> 839 LA Times: crime classifier 840 </a> 841 <div class="toc"></div> 842 843 844 845 </li> 846 847 </ol> 848 </div> 849 850 </li> 851 852 </ol> 853 854 <h4 id="howxaffectsy" class="sidebar-sticky"> 855 <a href="#howxaffectsy">How X affects Y</a> 856 </h4> 857 <ol class="menu menu-nav"> 858 859 <li class="menu-item"> 860 861 <a href="/regression/what-is-regression/"> 862 Finding relationships with regression 863 </a> 864 <div class="toc"></div> 865 866 867 868 </li> 869 870 <li class="menu-item"> 871 872 873 874 <input id="accordion-linearregression" type="checkbox" name="accordion-checkbox" hidden=""> 875 <label class="accordion-header c-hand" for="accordion-linearregression"> 876 <i class="icon icon-arrow-right mr-1"></i> 877 Linear Regression 878 </label> 879 <div class="accordion-body"> 880 <ol class="menu menu-nav"> 881 882 <li class="menu-item"> 883 884 <a href="/regression/linear-regression-quickstart/"> 885 Linear regression (Quickstart) 886 </a> 887 <div class="toc"></div> 888 889 890 891 </li> 892 893 <li class="menu-item"> 894 895 <a href="/regression/linear-regression/"> 896 Linear regression for humans 897 </a> 898 <div class="toc"></div> 899 900 901 902 </li> 903 904 <li class="menu-item"> 905 906 <a href="/regression/linear-regression-part-two/"> 907 Putting regression to use 908 </a> 909 <div class="toc"></div> 910 911 912 913 </li> 914 915 <li class="menu-item"> 916 917 <a href="/regression/linear-regression-evaluation/"> 918 Evaluating regressions 919 </a> 920 <div class="toc"></div> 921 922 923 924 </li> 925 926 <li class="menu-item"> 927 928 <a href="/ap-regression-unemployment/simple-regression-with-census-data-statsmodels-with-formulas/"> 929 Associated Press: Life expectancy and unemployment 930 </a> 931 <div class="toc"></div> 932 933 934 935 </li> 936 937 </ol> 938 </div> 939 940 </li> 941 942 <li class="menu-item"> 943 944 945 946 <input id="accordion-logisticregression" type="checkbox" name="accordion-checkbox" hidden=""> 947 <label class="accordion-header c-hand" for="accordion-logisticregression"> 948 <i class="icon icon-arrow-right mr-1"></i> 949 Logistic Regression 950 </label> 951 <div class="accordion-body"> 952 <ol class="menu menu-nav"> 953 954 <li class="menu-item"> 955 956 <a href="/regression/logistic-regression-quickstart/"> 957 Logistic regression (Quickstart) 958 </a> 959 <div class="toc"></div> 960 961 962 963 </li> 964 965 <li class="menu-item"> 966 967 <a href="/regression/logistic-regression/"> 968 Logistic regression for humans 969 </a> 970 <div class="toc"></div> 971 972 973 974 </li> 975 976 <li class="menu-item"> 977 978 <a href="/regression/logistic-regression-part-two/"> 979 More complex logistic regressions 980 </a> 981 <div class="toc"></div> 982 983 984 985 </li> 986 987 <li class="menu-item"> 988 989 <a href="/regression/evaluating-logistic-regressions/"> 990 Evaluating logistic regressions 991 </a> 992 <div class="toc"></div> 993 994 995 996 </li> 997 998 <li class="menu-item"> 999 1000 <a href="/boston-globe-tickets/boston-globe-ticketing-regression/"> 1001 Boston Globe: Speeding tickets 1002 </a> 1003 <div class="toc"></div> 1004 1005 1006 1007 </li> 1008 1009 <li class="menu-item"> 1010 1011 <a href="/apm-reports-jury-bias/in-the-dark-alternative-formula-methods/"> 1012 APM Reports: Jury selection 1013 </a> 1014 <div class="toc"></div> 1015 1016 1017 1018 </li> 1019 1020 </ol> 1021 </div> 1022 1023 </li> 1024 1025 </ol> 1026 1027 <h4 id="pythondatasciencereference" class="sidebar-sticky"> 1028 <a href="#pythondatasciencereference">Python data science reference</a> 1029 </h4> 1030 <ol class="menu menu-nav"> 1031 1032 <li class="menu-item"> 1033 1034 <a href="/reference/"> 1035 Introduction 1036 </a> 1037 <div class="toc"></div> 1038 1039 1040 1041 </li> 1042 1043 <li class="menu-item"> 1044 1045 <a href="/reference/vectorizing/"> 1046 Vectorizing 1047 </a> 1048 <div class="toc"></div> 1049 1050 1051 1052 </li> 1053 1054 <li class="menu-item"> 1055 1056 <a href="/reference/text-analysis/"> 1057 Text Analysis 1058 </a> 1059 <div class="toc"></div> 1060 1061 1062 1063 </li> 1064 1065 <li class="menu-item"> 1066 1067 <a href="/reference/regression/"> 1068 Regression 1069 </a> 1070 <div class="toc"></div> 1071 1072 1073 1074 </li> 1075 1076 <li class="menu-item"> 1077 1078 <a href="/reference/classification/"> 1079 Classification 1080 </a> 1081 <div class="toc"></div> 1082 1083 1084 1085 </li> 1086 1087 </ol> 1088 1089 <h4 id="allprojects" class="sidebar-sticky"> 1090 <a href="#allprojects">All Projects</a> 1091 </h4> 1092 <ol class="menu menu-nav"> 1093 1094 <li class="menu-item"> 1095 1096 <a href="/projects/"> 1097 Project Summaries 1098 </a> 1099 <div class="toc"></div> 1100 1101 1102 1103 </li> 1104 1105 <li class="menu-item"> 1106 1107 1108 1109 <input id="accordion-strongnytstrongtakataairbagsearch" type="checkbox" name="accordion-checkbox" hidden=""> 1110 <label class="accordion-header c-hand" for="accordion-strongnytstrongtakataairbagsearch"> 1111 <i class="icon icon-arrow-right mr-1"></i> 1112 <strong>NYT:</strong> Takata airbag search 1113 </label> 1114 <div class="accordion-body"> 1115 <ol class="menu menu-nav"> 1116 1117 <li class="menu-item"> 1118 1119 <a href="/nyt-takata-airbags/"> 1120 Summary 1121 </a> 1122 <div class="toc"></div> 1123 1124 1125 1126 </li> 1127 1128 <li class="menu-item"> 1129 1130 <a href="/nyt-takata-airbags/airbag-classifier-search-binary/"> 1131 A simplistic reproduction of the NYT's research using logistic regression 1132 </a> 1133 <div class="toc"></div> 1134 1135 1136 1137 </li> 1138 1139 <li class="menu-item"> 1140 1141 <a href="/nyt-takata-airbags/airbag-classifier-search-decision-tree/"> 1142 A decision-tree reproduction of the NYT's research 1143 </a> 1144 <div class="toc"></div> 1145 1146 1147 1148 </li> 1149 1150 <li class="menu-item"> 1151 1152 <a href="/nyt-takata-airbags/airbag-classifier-search-countvectorizer/"> 1153 Combining a text vectorizer and a classifier to track down suspicious complaints 1154 </a> 1155 <div class="toc"></div> 1156 1157 1158 1159 </li> 1160 1161 </ol> 1162 </div> 1163 1164 </li> 1165 1166 <li class="menu-item"> 1167 1168 1169 1170 <input id="accordion-stronglatimesstrongcrimeclassification" type="checkbox" name="accordion-checkbox" hidden=""> 1171 <label class="accordion-header c-hand" for="accordion-stronglatimesstrongcrimeclassification"> 1172 <i class="icon icon-arrow-right mr-1"></i> 1173 <strong>LA Times:</strong> Crime classification 1174 </label> 1175 <div class="accordion-body"> 1176 <ol class="menu menu-nav"> 1177 1178 <li class="menu-item"> 1179 1180 <a href="/latimes-crime-classification/"> 1181 Summary 1182 </a> 1183 <div class="toc"></div> 1184 1185 1186 1187 </li> 1188 1189 <li class="menu-item"> 1190 1191 <a href="/latimes-crime-classification/using-a-classifier-to-find-misclassified-crimes/"> 1192 Predicting downgraded assaults with machine learning 1193 </a> 1194 <div class="toc"></div> 1195 1196 1197 1198 </li> 1199 1200 <li class="menu-item"> 1201 1202 <a href="/latimes-crime-classification/inspecting-classifications/"> 1203 Taking a closer look at our classifier and its misclassifications 1204 </a> 1205 <div class="toc"></div> 1206 1207 1208 1209 </li> 1210 1211 <li class="menu-item"> 1212 1213 <a href="/latimes-crime-classification/trying-out-different-classifiers/"> 1214 Trying out and combining different classifiers 1215 </a> 1216 <div class="toc"></div> 1217 1218 1219 1220 </li> 1221 1222 </ol> 1223 </div> 1224 1225 </li> 1226 1227 <li class="menu-item"> 1228 1229 1230 1231 <input id="accordion-strongcaixinstrongmuseumnames" type="checkbox" name="accordion-checkbox" hidden=""> 1232 <label class="accordion-header c-hand" for="accordion-strongcaixinstrongmuseumnames"> 1233 <i class="icon icon-arrow-right mr-1"></i> 1234 <strong>Caixin:</strong> Museum names 1235 </label> 1236 <div class="accordion-body"> 1237 <ol class="menu menu-nav"> 1238 1239 <li class="menu-item"> 1240 1241 <a href="/caixin-museum-word-count/"> 1242 Summary 1243 </a> 1244 <div class="toc"></div> 1245 1246 1247 1248 </li> 1249 1250 <li class="menu-item"> 1251 1252 <a href="/caixin-museum-word-count/chinese-museum-dataset-cleanup/"> 1253 Chinese museum dataset cleanup 1254 </a> 1255 <div class="toc"></div> 1256 1257 1258 1259 </li> 1260 1261 <li class="menu-item"> 1262 1263 <a href="/caixin-museum-word-count/chinese-museums-per-capita-analysis/"> 1264 Chinese museums per capita analysis 1265 </a> 1266 <div class="toc"></div> 1267 1268 1269 1270 </li> 1271 1272 <li class="menu-item"> 1273 1274 <a href="/caixin-museum-word-count/counting-words-in-chinese-museum-names/"> 1275 Counting words in Chinese museum names 1276 </a> 1277 <div class="toc"></div> 1278 1279 1280 1281 </li> 1282 1283 </ol> 1284 </div> 1285 1286 </li> 1287 1288 <li class="menu-item"> 1289 1290 1291 1292 <input id="accordion-strongwapostrongrandomchatappsafety" type="checkbox" name="accordion-checkbox" hidden=""> 1293 <label class="accordion-header c-hand" for="accordion-strongwapostrongrandomchatappsafety"> 1294 <i class="icon icon-arrow-right mr-1"></i> 1295 <strong>WaPo:</strong> Random chat app safety 1296 </label> 1297 <div class="accordion-body"> 1298 <ol class="menu menu-nav"> 1299 1300 <li class="menu-item"> 1301 1302 <a href="/wapo-app-reviews/"> 1303 Summary 1304 </a> 1305 <div class="toc"></div> 1306 1307 1308 1309 </li> 1310 1311 <li class="menu-item"> 1312 1313 <a href="/wapo-app-reviews/scrape-app-store-reviews/"> 1314 Scrape and combine app store reviews 1315 </a> 1316 <div class="toc"></div> 1317 1318 1319 1320 </li> 1321 1322 <li class="menu-item"> 1323 1324 <a href="/wapo-app-reviews/predict-reviews/"> 1325 Build a classifier to detect reviews about bad behavior 1326 </a> 1327 <div class="toc"></div> 1328 1329 1330 1331 </li> 1332 1333 </ol> 1334 </div> 1335 1336 </li> 1337 1338 <li class="menu-item"> 1339 1340 <a href="/ajc-doctors-abuse/"> 1341 <strong>AJC:</strong> Doctors and sex abuse 1342 </a> 1343 <div class="toc"></div> 1344 1345 1346 1347 </li> 1348 1349 <li class="menu-item"> 1350 1351 1352 1353 <input id="accordion-strongtheupshotstrongtrumpspeeches" type="checkbox" name="accordion-checkbox" hidden=""> 1354 <label class="accordion-header c-hand" for="accordion-strongtheupshotstrongtrumpspeeches"> 1355 <i class="icon icon-arrow-right mr-1"></i> 1356 <strong>The UpShot:</strong> Trump speeches 1357 </label> 1358 <div class="accordion-body"> 1359 <ol class="menu menu-nav"> 1360 1361 <li class="menu-item"> 1362 1363 <a href="/upshot-trump-emolex/"> 1364 Summary 1365 </a> 1366 <div class="toc"></div> 1367 1368 1369 1370 </li> 1371 1372 <li class="menu-item"> 1373 1374 <a href="/upshot-trump-emolex/nrc-emotional-lexicon/"> 1375 An introduction to the NRC Emotional Lexicon 1376 </a> 1377 <div class="toc"></div> 1378 1379 1380 1381 </li> 1382 1383 <li class="menu-item"> 1384 1385 <a href="/upshot-trump-emolex/trump-vs-state-of-the-union-addresses/"> 1386 Reproducing The UpShot's Trump State of the Union visualization 1387 </a> 1388 <div class="toc"></div> 1389 1390 1391 1392 </li> 1393 1394 </ol> 1395 </div> 1396 1397 </li> 1398 1399 <li class="menu-item"> 1400 1401 1402 1403 <input id="accordion-strongusatodaystrongmodellegislation" type="checkbox" name="accordion-checkbox" hidden=""> 1404 <label class="accordion-header c-hand" for="accordion-strongusatodaystrongmodellegislation"> 1405 <i class="icon icon-arrow-right mr-1"></i> 1406 <strong>USA Today:</strong> Model legislation 1407 </label> 1408 <div class="accordion-body"> 1409 <ol class="menu menu-nav"> 1410 1411 <li class="menu-item"> 1412 1413 <a href="/azcentral-text-reuse-model-legislation/"> 1414 Summary 1415 </a> 1416 <div class="toc"></div> 1417 1418 1419 1420 </li> 1421 1422 <li class="menu-item"> 1423 1424 <a href="/azcentral-text-reuse-model-legislation/01-downloading-one-million-pieces-of-legislation-from-legiscan/"> 1425 Downloading one million pieces of legislation from LegiScan 1426 </a> 1427 <div class="toc"></div> 1428 1429 1430 1431 </li> 1432 1433 <li class="menu-item"> 1434 1435 <a href="/azcentral-text-reuse-model-legislation/02-taking-a-mill
1435ion-pieces-of-legislation-from-a-csv-and-inserting-them-into-postgres/"> 1436 Taking a million pieces of legislation from a CSV and inserting them into Postgres 1437 </a> 1438 <div class="toc"></div> 1439 1440 1441 1442 </li> 1443 1444 <li class="menu-item"> 1445 1446 <a href="/azcentral-text-reuse-model-legislation/03-download-word-pdf-and-html-content-and-process-it-into-text-with-tika/"> 1447 Download Word, PDF and HTML content and process it into text with Tika 1448 </a> 1449 <div class="toc"></div> 1450 1451 1452 1453 </li> 1454 1455 <li class="menu-item"> 1456 1457 <a href="/azcentral-text-reuse-model-legislation/04-import-content-into-solr-for-advanced-text-searching/"> 1458 Import content into Solr for advanced text searching 1459 </a> 1460 <div class="toc"></div> 1461 1462 1463 1464 </li> 1465 1466 <li class="menu-item"> 1467 1468 <a href="/azcentral-text-reuse-model-legislation/05-checking-for-legislative-text-reuse-using-python-solr-and-ngrams/"> 1469 Checking for legislative text reuse using Python, Solr, and ngrams 1470 </a> 1471 <div class="toc"></div> 1472 1473 1474 1475 </li> 1476 1477 <li class="menu-item"> 1478 1479 <a href="/azcentral-text-reuse-model-legislation/05-checking-for-legislative-text-reuse-using-python-solr-and-simple-text-search/"> 1480 Checking for legislative text reuse using Python, Solr, and simple text search 1481 </a> 1482 <div class="toc"></div> 1483 1484 1485 1486 </li> 1487 1488 <li class="menu-item"> 1489 1490 <a href="/azcentral-text-reuse-model-legislation/06-search-for-model-legislation-in-over-one-million-bills-using-postgres-and-solr/"> 1491 Search for model legislation in over one million bills using Postgres and Solr 1492 </a> 1493 <div class="toc"></div> 1494 1495 1496 1497 </li> 1498 1499 <li class="menu-item"> 1500 1501 <a href="/azcentral-text-reuse-model-legislation/using-topic-modeling-to-categorize-legislation/"> 1502 Using topic modeling to categorize legislation 1503 </a> 1504 <div class="toc"></div> 1505 1506 1507 1508 </li> 1509 1510 </ol> 1511 </div> 1512 1513 </li> 1514 1515 <li class="menu-item"> 1516 1517 <a href="/fcc-comments/"> 1518 FCC comment bots 1519 </a> 1520 <div class="toc"></div> 1521 1522 1523 1524 </li> 1525 1526 <li class="menu-item"> 1527 1528 1529 1530 <input id="accordion-strongbloombergstrongdemocraticcandidatetweets" type="checkbox" name="accordion-checkbox" hidden=""> 1531 <label class="accordion-header c-hand" for="accordion-strongbloombergstrongdemocraticcandidatetweets"> 1532 <i class="icon icon-arrow-right mr-1"></i> 1533 <strong>Bloomberg:</strong> Democratic Candidate Tweets 1534 </label> 1535 <div class="accordion-body"> 1536 <ol class="menu menu-nav"> 1537 1538 <li class="menu-item"> 1539 1540 <a href="/bloomberg-tweet-topics/"> 1541 Summary 1542 </a> 1543 <div class="toc"></div> 1544 1545 1546 1547 </li> 1548 1549 <li class="menu-item"> 1550 1551 <a href="/bloomberg-tweet-topics/scrape-tweets-from-presidential-primary-candidates/"> 1552 Downloading all 2019 tweets from Democratic presidential candidates 1553 </a> 1554 <div class="toc"></div> 1555 1556 1557 1558 </li> 1559 1560 <li class="menu-item"> 1561 1562 <a href="/bloomberg-tweet-topics/topic-modeling-for-tweets/"> 1563 Using topic modeling to analyze presidential candidate tweets 1564 </a> 1565 <div class="toc"></div> 1566 1567 1568 1569 </li> 1570 1571 <li class="menu-item"> 1572 1573 <a href="/bloomberg-tweet-topics/assigning-categories-to-text-using-keyword-matching/"> 1574 Assigning categories to tweets using keyword matching 1575 </a> 1576 <div class="toc"></div> 1577 1578 1579 1580 </li> 1581 1582 <li class="menu-item"> 1583 1584 <a href="/bloomberg-tweet-topics/building-streamgraphs-from-candidate-tweets/"> 1585 Building streamgraphs from categorized and dated datasets 1586 </a> 1587 <div class="toc"></div> 1588 1589 1590 1591 </li> 1592 1593 </ol> 1594 </div> 1595 1596 </li> 1597 1598 <li class="menu-item"> 1599 1600 <a href="/nyt-trump-tweets/"> 1601 <strong>NYT:</strong> Trump tweets 1602 </a> 1603 <div class="toc"></div> 1604 1605 1606 1607 </li> 1608 1609 <li class="menu-item"> 1610 1611 1612 1613 <input id="accordion-strongapstronglifeexpectancy" type="checkbox" name="accordion-checkbox" hidden=""> 1614 <label class="accordion-header c-hand" for="accordion-strongapstronglifeexpectancy"> 1615 <i class="icon icon-arrow-right mr-1"></i> 1616 <strong>AP:</strong> Life expectancy 1617 </label> 1618 <div class="accordion-body"> 1619 <ol class="menu menu-nav"> 1620 1621 <li class="menu-item"> 1622 1623 <a href="/ap-regression-unemployment/"> 1624 Summary 1625 </a> 1626 <div class="toc"></div> 1627 1628 1629 1630 </li> 1631 1632 <li class="menu-item"> 1633 1634 <a href="/ap-regression-unemployment/simple-regression-with-census-data-statsmodels-with-formulas/"> 1635 Simple logistic regression using statsmodels (formula version) 1636 </a> 1637 <div class="toc"></div> 1638 1639 1640 1641 </li> 1642 1643 <li class="menu-item"> 1644 1645 <a href="/ap-regression-unemployment/simple-regression-with-census-data-statsmodels-with-dataframes/"> 1646 Simple logistic regression using statsmodels (dataframes version) 1647 </a> 1648 <div class="toc"></div> 1649 1650 1651 1652 </li> 1653 1654 </ol> 1655 </div> 1656 1657 </li> 1658 1659 <li class="menu-item"> 1660
1661 <a href="/fivethirtyeight-p-hacking/"> 1662 <strong>FiveThirtyEight:</strong> P-values 1663 </a> 1664 <div class="toc"></div> 1665 1666 1667 1668 </li> 1669 1670 <li class="menu-item"> 1671 1672 1673 1674 <input id="accordion-strongmilwaukeejournalsentinelstrongpotholes" type="checkbox" name="accordion-checkbox" hidden=""> 1675 <label class="accordion-header c-hand" for="accordion-strongmilwaukeejournalsentinelstrongpotholes"> 1676 <i class="icon icon-arrow-right mr-1"></i> 1677 <strong>Milwaukee Journal-Sentinel:</strong> Potholes 1678 </label> 1679 <div class="accordion-body"> 1680 <ol class="menu menu-nav"> 1681 1682 <li class="menu-item"> 1683 1684 <a href="/milwaukee-potholes/"> 1685 Summary 1686 </a> 1687 <div class="toc"></div> 1688 1689 1690 1691 </li> 1692 1693 <li class="menu-item"> 1694 1695 <a href="/milwaukee-potholes/milwaukee-journal-sentinel-and-potholes-full-walkthrough/"> 1696 Pothole geographic analysis and linear regression, complete walkthrough 1697 </a> 1698 <div class="toc"></div> 1699 1700 1701 1702 </li> 1703 1704 <li class="menu-item"> 1705 1706 <a href="/milwaukee-potholes/milwaukee-journal-sentinel-and-potholes-without-merging/"> 1707 Pothole demographics linear regression, no spatial analysis 1708 </a> 1709 <div class="toc"></div> 1710 1711 1712 1713 </li> 1714 1715 </ol> 1716 </div> 1717 1718 </li> 1719 1720 <li class="menu-item"> 1721 1722 1723 1724 <input id="accordion-strongdallasmorningnewsstrongcheatingschools" type="checkbox" name="accordion-checkbox" hidden=""> 1725 <label class="accordion-header c-hand" for="accordion-strongdallasmorningnewsstrongcheatingschools"> 1726 <i class="icon icon-arrow-right mr-1"></i> 1727 <strong>Dallas Morning News:</strong> Cheating schools 1728 </label> 1729 <div class="accordion-body"> 1730 <ol class="menu menu-nav"> 1731 1732 <li class="menu-item"> 1733 1734 <a href="/dmn-texas-school-cheating/"> 1735 Summary 1736 </a> 1737 <div class="toc"></div> 1738 1739 1740 1741 </li> 1742 1743 <li class="menu-item"> 1744 1745 <a href="/dmn-texas-school-cheating/texas-school-cheating-finding-outliers-with-standard-deviation-and-regression/"> 1746 Finding outliers with standard deviation and regression 1747 </a> 1748 <div class="toc"></div> 1749 1750 1751 1752 </li> 1753 1754 <li class="menu-item"> 1755 1756 <a href="/dmn-texas-school-cheating/texas-school-cheating-finding-outliers-with-regression-residuals/"> 1757 Finding outliers with regression residuals (short version) 1758 </a> 1759 <div class="toc"></div> 1760 1761 1762 1763 </li> 1764 1765 <li class="menu-item"> 1766 1767 <a href="/dmn-texas-school-cheating/texas-school-cheating-graph-reproductions/"> 1768 Reproducing the graphics from The Dallas Morning News piece 1769 </a> 1770 <div class="toc"></div> 1771 1772 1773 1774 </li> 1775 1776 </ol> 1777 </div> 1778 1779 </li> 1780 1781 <li class="menu-item"> 1782 1783 1784 1785 <input id="accordion-strongtampabaytimesstrongfailurefactories" type="checkbox" name="accordion-checkbox" hidden=""> 1786 <label class="accordion-header c-hand" for="accordion-strongtampabaytimesstrongfailurefactories"> 1787 <i class="icon icon-arrow-right mr-1"></i> 1788 <strong>Tampa Bay Times:</strong> Failure factories 1789 </label> 1790 <div class="accordion-body"> 1791 <ol class="menu menu-nav"> 1792 1793 <li class="menu-item"> 1794 1795 <a href="/tampa-bay-times-schools/"> 1796 Summary 1797 </a> 1798 <div class="toc"></div> 1799 1800 1801 1802 </li> 1803 1804 <li class="menu-item"> 1805 1806 <a href="/tampa-bay-times-schools/linear-regression-on-florida-schools/"> 1807 Linear regression on Florida schools, complete walkthrough 1808 </a> 1809 <div class="toc"></div> 1810 1811 1812 1813 </li> 1814 1815 <li class="menu-item"> 1816 1817 <a href="/tampa-bay-times-schools/linear-regression-on-florida-schools-no-cleaning/"> 1818 Linear regression on Florida schools, no cleaning 1819 </a> 1820 <div class="toc"></div> 1821 1822 1823 1824 </li> 1825 1826 </ol> 1827 </div> 1828 1829 </li> 1830 1831 <li class="menu-item"> 1832 1833 1834 1835 <input id="accordion-caraccidentsandcarweight" type="checkbox" name="accordion-checkbox" hidden=""> 1836 <label class="accordion-header c-hand" for="accordion-caraccidentsandcarweight"> 1837 <i class="icon icon-arrow-right mr-1"></i> 1838 Car accidents and car weight 1839 </label> 1840 <div class="accordion-body"> 1841 <ol class="menu menu-nav"> 1842 1843 <li class="menu-item"> 1844 1845 <a href="/car-crashes-weight-regression/"> 1846 Summary 1847 </a> 1848 <div class="toc"></div> 1849 1850 1851 1852 </li> 1853 1854 <li class="menu-item"> 1855 1856 <a href="/car-crashes-weight-regression/car-crashes-feature-selection-and-engineering/"> 1857 Feature selection and engineering
1858 </a> 1859 <div class="toc"></div> 1860 1861 1862 1863 </li> 1864 1865 <li class="menu-item"> 1866 1867 <a href="/car-crashes-weight-regression/01-combine-excel-files-across-multiple-sheets-and-save-as-csv-files/"> 1868 Combine Excel files across multiple sheets and save as CSV files 1869 </a> 1870 <div class="toc"></div> 1871 1872 1873 1874 </li> 1875 1876 <li class="menu-item"> 1877 1878 <a href="/car-crashes-weight-regression/02-create-make-model-weights-csv/"> 1879 Create make model weights csv 1880 </a> 1881 <div class="toc"></div> 1882 1883 1884 1885 </li> 1886 1887 <li class="menu-item"> 1888 1889 <a href="/car-crashes-weight-regression/03-find-car-data-from-vins/"> 1890 Find car data from VINs 1891 </a> 1892 <div class="toc"></div> 1893 1894 1895 1896 </li> 1897 1898 <li class="menu-item"> 1899 1900 <a href="/car-crashes-weight-regression/04-combine-vins-and-weights/"> 1901 Combine VINs and weights 1902 </a> 1903 <div class="toc"></div> 1904 1905 1906 1907 </li> 1908 1909 <li class="menu-item"> 1910 1911 <a href="/car-crashes-weight-regression/05-clean-combine-and-filter-data/"> 1912 Clean combine and filter data 1913 </a> 1914 <div class="toc"></div> 1915 1916 1917 1918 </li> 1919 1920 </ol> 1921 </div> 1922 1923 </li> 1924 1925 <li class="menu-item"> 1926 1927 <a href="/propublica-opportunity-gap/"> 1928 <strong>ProPublica:</strong> Opportunity Gap 1929 </a> 1930 <div class="toc"></div> 1931 1932 1933 1934 </li> 1935 1936 <li class="menu-item"> 1937 1938 1939 1940 <input id="accordion-strongbostonglobestrongticketingbias" type="checkbox" name="accordion-checkbox" hidden=""> 1941 <label class="accordion-header c-hand" for="accordion-strongbostonglobestrongticketingbias"> 1942 <i class="icon icon-arrow-right mr-1"></i> 1943 <strong>Boston Globe:</strong> Ticketing bias 1944 </label> 1945 <div class="accordion-body"> 1946 <ol class="menu menu-nav"> 1947 1948 <li class="menu-item"> 1949 1950 <a href="/boston-globe-tickets/"> 1951 Summary 1952 </a> 1953 <div class="toc"></div> 1954 1955 1956 1957 </li> 1958 1959 <li class="menu-item"> 1960 1961 <a href="/boston-globe-tickets/boston-globe-ticketing-regression/"> 1962 Logistic regression for speeding tickets 1963 </a> 1964 <div class="toc"></div> 1965 1966 1967 1968 </li> 1969 1970 </ol> 1971 </div> 1972 1973 </li> 1974 1975 <li class="menu-item"> 1976 1977 <a href="/stanford-open-policing/"> 1978 <strong>Stanford:</strong> Open Policing Data 1979 </a> 1980 <div class="toc"></div> 1981 1982 1983 1984 </li> 1985 1986 <li class="menu-item"> 1987 1988 1989 1990 <input id="accordion-strongbuzzfeedstrongsurveillanceplanes" type="checkbox" name="accordion-checkbox" hidden=""> 1991 <label class="accordion-header c-hand" for="accordion-strongbuzzfeedstrongsurveillanceplanes"> 1992 <i class="icon icon-arrow-right mr-1"></i> 1993 <strong>BuzzFeed:</strong> Surveillance planes 1994 </label> 1995 <div class="accordion-body"> 1996 <ol class="menu menu-nav"> 1997 1998 <li class="menu-item"> 1999 2000 <a href="/buzzfeed-spy-planes/"> 2001 Summary 2002 </a> 2003 <div class="toc"></div> 2004 2005 2006 2007 </li> 2008 2009 <li class="menu-item"> 2010 2011 <a href="/buzzfeed-spy-planes/feature-engineering-buzzfeed-spy-planes/"> 2012 Feature engineering - BuzzFeed spy planes 2013 </a> 2014 <div class="toc"></div> 2015 2016 2017 2018 </li> 2019 2020 <li class="menu-item"> 2021 2022 <a href="/buzzfeed-spy-planes/drawing-flight-paths-on-maps-with-cartopy/"> 2023 Drawing flight paths on maps with cartopy 2024 </a> 2025 <div class="toc"></div> 2026 2027 2028 2029 </li> 2030 2031 <li class="menu-item"> 2032 2033 <a href="/buzzfeed-spy-planes/buzzfeed-surveillance-planes-random-forests/"> 2034 Finding surveillance planes using random forests 2035 </a> 2036 <div class="toc"></div> 2037 2038 2039 2040 </li> 2041 2042 </ol> 2043 </div> 2044 2045 </li> 2046 2047 <li class="menu-item"> 2048 2049 2050 2051 <input id="accordion-strongrevealstrongmortgagelendingbias" type="checkbox" name="accordion-checkbox" hidden=""> 2052 <label class="accordion-header c-hand" for="accordion-strongrevealstrongmortgagelendingbias"> 2053 <i class="icon icon-arrow-right mr-1"></i> 2054 <strong>Reveal:</strong> Mortgage lending bias 2055 </label> 2056 <div class="accordion-body"> 2057 <ol class="menu menu-nav"> 2058 2059 <li class="menu-item"> 2060 2061 <a href="/reveal-mortgages/"> 2062 Summary 2063 </a> 2064 <div class="toc"></div> 2065 2066 2067 2068 </li> 2069 2070 <li class="menu-item"> 2071 2072 <a href="/reveal-mortgages/reveal-mortgage-analysis-cleaning-and-combining-data/"> 2073 Cleaning and combining data for the Reveal Mortgage Analysis 2074 </a> 2075 <div class="toc"></div> 2076 2077 2078 2079 </li> 2080 2081 <li class="menu-item"> 2082 2083 <a href="/reveal-mortgages/reveal-mortgage-analysis-wild-formulas-in-statsmodels-using-patsy-short-version/"> 2084 Wild formulas in statsmodels using Patsy (short version) 2085 </a> 2086 <div class="toc"></div> 2087 2088 2089 2090 </li> 2091 2092 <li class="menu-item"> 2093 2094 <a href="/reveal-mortgages/reveal-mortgage-analysis-logistic-regression-using-statsmodels-formulas/"> 2095 Reveal Mortgage Analysis - Logistic Regression using statsmodels formulas 2096 </a> 2097 <div class="toc"></div> 2098 2099 2100 2101 </li> 2102 2103 <li class="menu-item"> 2104 2105 <a href="/reveal-mortgages/reveal-mortgage-analysis-logistic-regression/"> 2106 Reveal Mortgage Analysis - Logistic Regression 2107 </a> 2108 <div class="toc"></div> 2109 2110 2111 2112 </li> 2113 2114 </ol> 2115 </div> 2116 2117 </li> 2118 2119 <li class="menu-item"> 2120 2121 2122 2123 <input id="accordion-strongapmreportsstrongjuryselectionbias" type="checkbox" name="accordion-checkbox" hidden=""> 2124 <label class="accordion-header c-hand" for="accordion-strongapmreportsstrongjuryselectionbias"> 2125 <i class="icon icon-arrow-right mr-1"></i> 2126 <strong>APM Reports:</strong> Jury selection bias 2127 </label> 2128 <div class="accordion-body"> 2129 <ol class="menu menu-nav"> 2130 2131 <li class="menu-item"> 2132 2133 <a href="/apm-reports-jury-bias/"> 2134 Summary 2135 </a> 2136 <div class="toc"></div> 2137 2138 2139 2140 </li> 2141 2142 <li class="menu-item"> 2143 2144 <a href="/apm-reports-jury-bias/in-the-dark-combining-datasets-and-cleaning-the-data/"> 2145 Combining and cleaning the initial dataset 2146 </a> 2147 <div class="toc"></div> 2148 2149 2150 2151 </li> 2152 2153 <li class="menu-item"> 2154 2155 <a href="/apm-reports-jury-bias/in-the-dark-feature-selection-with-p-values/"> 2156 Picking what matters and what doesn't in a regression 2157 </a> 2158 <div class="toc"></div> 2159 2160 2161 2162 </li> 2163 2164 <li class="menu-item"> 2165 2166 <a href="/apm-reports-jury-bias/in-the-dark-jury-selection-regression-walkthrough/"> 2167 Analyzing data using statsmodels formulas 2168 </a> 2169 <div class="toc"></div> 2170 2171 2172 2173 </li> 2174 2175 <li class="menu-item"> 2176 2177 <a href="/apm-reports-jury-bias/in-the-dark-alternative-formula-methods/"> 2178 Alternative techniques with statsmodels formulas 2179 </a> 2180 <div class="toc"></div> 2181 2182 2183 2184 </li> 2185 2186 </ol> 2187 </div> 2188 2189 </li> 2190 2191 <li class="menu-item"> 2192 2193 2194 2195 <input id="accordion-strongreutersstrongasylumdenials" type="checkbox" name="accordion-checkbox" hidden=""> 2196 <label class="accordion-header c-hand" for="accordion-strongreutersstrongasylumdenials"> 2197 <i class="icon icon-arrow-right mr-1"></i> 2198 <strong>Reuters:</strong> Asylum denials 2199 </label> 2200 <div class="accordion-body"> 2201 <ol class="menu menu-nav"> 2202 2203 <li class="menu-item"> 2204 2205 <a href="/reuters-asylum/"> 2206 Summary 2207 </a> 2208 <div class="toc"></div> 2209 2210 2211 2212 </li> 2213 2214 <li class="menu-item"> 2215 2216 <a href="/reuters-asylum/cleaning-the-eoir-immigration-court-dataset/"> 2217 Preparing the EOIR immigration court data for analysis 2218 </a> 2219 <div class="toc"></div> 2220 2221 2222 2223 </li> 2224 2225 <li class="menu-item"> 2226 2227 <a href="/reuters-asylum/using-regression-to-analyze-asylum-cases/"> 2228 How nationality and judges affect your chance of asylum in immigration court 2229 </a> 2230 <div class="toc"></div> 2231 2232 2233 2234 </li> 2235 2236 </ol> 2237 </div> 2238 2239 </li> 2240 2241 <li class="menu-item"> 2242 2243 <a href="/propublica-pardons/"> 2244 <strong>ProPublica:</strong> Presidential pardons 2245 </a> 2246 <div class="toc"></div> 2247 2248 2249 2250 </li> 2251 2252 <li class="menu-item"> 2253 2254 2255 2256 <input id="accordion-strongpropublicastrongcriminalsentencing" t
2256ype="checkbox" name="accordion-checkbox" hidden=""> 2257 <label class="accordion-header c-hand" for="accordion-strongpropublicastrongcriminalsentencing"> 2258 <i class="icon icon-arrow-right mr-1"></i> 2259 <strong>ProPublica:</strong> Criminal sentencing 2260 </label> 2261 <div class="accordion-body"> 2262 <ol class="menu menu-nav"> 2263 2264 <li class="menu-item"> 2265 2266 <a href="/propublica-criminal-sentencing/"> 2267 Summary 2268 </a> 2269 <div class="toc"></div> 2270 2271 2272 2273 </li> 2274 2275 <li class="menu-item"> 2276 2277 <a href="/propublica-criminal-sentencing/week-5-1-machine-bias-class/"> 2278 Breaking down machine bias 2279 </a> 2280 <div class="toc"></div> 2281 2282 2283 2284 </li> 2285 2286 </ol> 2287 </div> 2288 2289 </li> 2290 2291 <li class="menu-item"> 2292 2293 <a href="/foia-predictor/"> 2294 <strong>data.world:</strong> The FOIA Predictor 2295 </a> 2296 <div class="toc"></div> 2297 2298 2299 2300 </li> 2301 2302 </ol> 2303 2304 </div> 2305 </div> 2306</div> 2307 </div> 2308 </div> 2309 </div> 2310
2310<script src="/js/tocbot.js"></script>
2310 2311 2312
2312<script> 2313 // Add the div to hold the table of contents 2314 try { 2315 let toc = document.createElement("div") 2316 toc.setAttribute('class', 'js-toc') 2317 document.querySelector(".reading-options").parentNode.appendChild(toc) 2318 } catch (err) { 2319 2320 } 2321 2322 let slugSelector = "/wapo-app-reviews/" 2323 2324 let contentSelector = '.notebook' 2325 2326 </script>
2326 2327 2328
2328<script src="/js/main.js?v=1edd6b72f7227d977df2bf125e3b6e55"></script>
2328 2329 2330</body> 2331 2332</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.