{"product_id":"responsible-data-science-paperback-softback-9781119741756","title":"Responsible Data Science (Paperback \/ softback) 9781119741756","description":"\u003cfont face=\"Georgia\"\u003e\r\n\u003cp\u003e\u003cfont size=\"6\"\u003eResponsible Data Science\u003c\/font\u003e\u003cbr\u003e\r\n\r\n\r\n\r\n\r\n\r\n\u003c\/p\u003e\n\u003cp\u003e\u003cfont size=\"4\"\u003eGrant Fleming (Author), Peter C. Bruce (Author)\u003c\/font\u003e\u003c\/p\u003e\r\n\r\n\u003cp\u003e\u003cfont size=\"3\"\u003e9781119741756, Wiley\u003c\/font\u003e\u003c\/p\u003e\r\n\r\n\u003cp\u003e\u003cfont size=\"3\"\u003ePaperback \/ softback, published 24 June 2021\u003c\/font\u003e\u003c\/p\u003e\r\n\r\n\u003cp\u003e\u003cfont size=\"3\"\u003e304 pages\u003cbr\u003e23.1 x 18.5 x 1.8 cm, 0.522 kg\u003c\/font\u003e\u003c\/p\u003e\r\n\r\n\r\n\r\n\r\n\r\n\u003cp align=\"justify\"\u003e\u003cstrong\u003e\u003cfont size=\"3\"\u003e\u003cp\u003e\u003cb\u003eExplore the most serious prevalent ethical issues in data science with this insightful new resource\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003eThe increasing popularity of data science has resulted in numerous well-publicized cases of bias, injustice, and discrimination. The widespread deployment of “Black box” algorithms that are difficult or impossible to understand and explain, even for their developers, is a primary source of these unanticipated harms, making modern techniques and methods for manipulating large data sets seem sinister, even dangerous. When put in the hands of authoritarian governments, these algorithms have enabled suppression of political dissent and persecution of minorities. To prevent these harms, data scientists everywhere must come to understand how the algorithms that they build and deploy may harm certain groups or be unfair.\u003c\/p\u003e \u003cp\u003e\u003ci\u003eResponsible Data Science\u003c\/i\u003e delivers a comprehensive, practical treatment of how to implement data science solutions in an even-handed and ethical manner that minimizes the risk of undue harm to vulnerable members of society. Both data science practitioners and managers of analytics teams will learn how to:\u003c\/p\u003e \u003cul\u003e \u003cli\u003eImprove model transparency, even for black box models\u003c\/li\u003e \u003cli\u003eDiagnose bias and unfairness within models using multiple metrics\u003c\/li\u003e \u003cli\u003eAudit projects to ensure fairness and minimize the possibility of unintended harm\u003c\/li\u003e \u003c\/ul\u003e \u003cp\u003ePerfect for data science practitioners, \u003ci\u003eResponsible Data Science\u003c\/i\u003e will also earn a spot on the bookshelves of technically inclined managers, software developers, and statisticians.\u003c\/p\u003e\u003c\/font\u003e\u003c\/strong\u003e\u003c\/p\u003e\r\n\r\n\u003cp\u003e\u003cfont size=\"3\"\u003e\u003cp\u003eIntroduction xix\u003c\/p\u003e \u003cp\u003e\u003cb\u003ePart I Motivation for Ethical Data Science and Background Knowledge 1\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003e\u003cb\u003eChapter 1 Responsible Data Science 3\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003eThe Optum Disaster 4\u003c\/p\u003e \u003cp\u003eJekyll and Hyde 5\u003c\/p\u003e \u003cp\u003eEugenics 7\u003c\/p\u003e \u003cp\u003eGalton, Pearson, and Fisher 7\u003c\/p\u003e \u003cp\u003eTies between Eugenics and Statistics 7\u003c\/p\u003e \u003cp\u003eEthical Problems in Data Science Today 9\u003c\/p\u003e \u003cp\u003ePredictive Models 10\u003c\/p\u003e \u003cp\u003eFrom Explaining to Predicting 10\u003c\/p\u003e \u003cp\u003ePredictive Modeling 11\u003c\/p\u003e \u003cp\u003eSetting the Stage for Ethical Issues to Arise 12\u003c\/p\u003e \u003cp\u003eClassic Statistical Models 12\u003c\/p\u003e \u003cp\u003eBlack-Box Methods 14\u003c\/p\u003e \u003cp\u003eImportant Concepts in Predictive Modeling 19\u003c\/p\u003e \u003cp\u003eFeature Selection 19\u003c\/p\u003e \u003cp\u003eModel-Centric vs. Data-Centric Models 20\u003c\/p\u003e \u003cp\u003eHoldout Sample and Cross-Validation 20\u003c\/p\u003e \u003cp\u003eOverfitting 21\u003c\/p\u003e \u003cp\u003eUnsupervised Learning 22\u003c\/p\u003e \u003cp\u003eThe Ethical Challenge of Black Boxes 23\u003c\/p\u003e \u003cp\u003eTwo Opposing Forces 24\u003c\/p\u003e \u003cp\u003ePressure for More Powerful AI 24\u003c\/p\u003e \u003cp\u003ePublic Resistance and Anxiety 24\u003c\/p\u003e \u003cp\u003eSummary 25\u003c\/p\u003e \u003cp\u003e\u003cb\u003eChapter 2 Background: Modeling and the Black-Box Algorithm 27\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003eAssessing Model Performance 27\u003c\/p\u003e \u003cp\u003ePredicting Class Membership 28\u003c\/p\u003e \u003cp\u003eThe Rare Class Problem 28\u003c\/p\u003e \u003cp\u003eLift and Gains 28\u003c\/p\u003e \u003cp\u003eArea Under the Curve 29\u003c\/p\u003e \u003cp\u003eAUC vs. Lift (Gains) 31\u003c\/p\u003e \u003cp\u003ePredicting Numeric Values 32\u003c\/p\u003e \u003cp\u003eGoodness-of-Fit 32\u003c\/p\u003e \u003cp\u003eHoldout Sets and Cross-Validation 33\u003c\/p\u003e \u003cp\u003eOptimization and Loss Functions 34\u003c\/p\u003e \u003cp\u003eIntrinsically Interpretable Models vs. Black-Box Models 35\u003c\/p\u003e \u003cp\u003eEthical Challenges with Interpretable Models 38\u003c\/p\u003e \u003cp\u003eBlack-Box Models 39\u003c\/p\u003e \u003cp\u003eEnsembles 39\u003c\/p\u003e \u003cp\u003eNearest Neighbors 41\u003c\/p\u003e \u003cp\u003eClustering 41\u003c\/p\u003e \u003cp\u003eAssociation Rules 42\u003c\/p\u003e \u003cp\u003eCollaborative Filters 42\u003c\/p\u003e \u003cp\u003eArtificial Neural Nets and Deep Neural Nets 43\u003c\/p\u003e \u003cp\u003eProblems with Black-Box Predictive Models 45\u003c\/p\u003e \u003cp\u003eProblems with Unsupervised Algorithms 47\u003c\/p\u003e \u003cp\u003eSummary 48\u003c\/p\u003e \u003cp\u003e\u003cb\u003eChapter 3 The Ways AI Goes Wrong, and the Legal Implications 49\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003eAI and Intentional Consequences by Design 50\u003c\/p\u003e \u003cp\u003eDeepfakes 50\u003c\/p\u003e \u003cp\u003eSupporting State Surveillance and Suppression 51\u003c\/p\u003e \u003cp\u003eBehavioral Manipulation 52\u003c\/p\u003e \u003cp\u003eAutomated Testing to Fine-Tune Targeting 53\u003c\/p\u003e \u003cp\u003eAI and Unintended Consequences 55\u003c\/p\u003e \u003cp\u003eHealthcare 56\u003c\/p\u003e \u003cp\u003eFinance 57\u003c\/p\u003e \u003cp\u003eLaw Enforcement 58\u003c\/p\u003e \u003cp\u003eTechnology 60\u003c\/p\u003e \u003cp\u003eThe Legal and Regulatory Landscape around AI 61\u003c\/p\u003e \u003cp\u003eIgnorance Is No Defense: AI in the Context of Existing Law and Policy 63\u003c\/p\u003e \u003cp\u003eA Finger in the Dam: Data Rights, Data Privacy, and Consumer Protection Regulations 64\u003c\/p\u003e \u003cp\u003eTrends in Emerging Law and Policy Related to AI 66\u003c\/p\u003e \u003cp\u003eSummary 69\u003c\/p\u003e \u003cp\u003e\u003cb\u003ePart II The Ethical Data Science Process 71\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003e\u003cb\u003eChapter 4 The Responsible Data Science Framework 73\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003eWhy We Keep Building Harmful AI 74\u003c\/p\u003e \u003cp\u003eMisguided Need for Cutting-Edge Models 74\u003c\/p\u003e \u003cp\u003eExcessive Focus on Predictive Performance 74\u003c\/p\u003e \u003cp\u003eEase of Access and the Curse of Simplicity 76\u003c\/p\u003e \u003cp\u003eThe Common Cause 76\u003c\/p\u003e \u003cp\u003eThe Face Thieves 78\u003c\/p\u003e \u003cp\u003eAn Anatomy of Modeling Harms 79\u003c\/p\u003e \u003cp\u003eThe World: Context Matters for Modeling 80\u003c\/p\u003e \u003cp\u003eThe Data: Representation Is Everything 83\u003c\/p\u003e \u003cp\u003eThe Model: Garbage In, Danger Out 85\u003c\/p\u003e \u003cp\u003eModel Interpretability: Human Understanding for Superhuman Models 86\u003c\/p\u003e \u003cp\u003eEfforts Toward a More Responsible Data Science 89\u003c\/p\u003e \u003cp\u003ePrinciples Are the Focus 90\u003c\/p\u003e \u003cp\u003eNonmaleficence 90\u003c\/p\u003e \u003cp\u003eFairness 90\u003c\/p\u003e \u003cp\u003eTransparency 91\u003c\/p\u003e \u003cp\u003eAccountability 91\u003c\/p\u003e \u003cp\u003ePrivacy 92\u003c\/p\u003e \u003cp\u003eBridging the Gap Between Principles and Practice with the Responsible Data Science (RDS) Framework 92\u003c\/p\u003e \u003cp\u003eJustification 94\u003c\/p\u003e \u003cp\u003eCompilation 94\u003c\/p\u003e \u003cp\u003ePreparation 95\u003c\/p\u003e \u003cp\u003eModeling 96\u003c\/p\u003e \u003cp\u003eAuditing 96\u003c\/p\u003e \u003cp\u003eSummary 97\u003c\/p\u003e \u003cp\u003e\u003cb\u003eChapter 5 Model Interpretability: The What and the Why 99\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003eThe Sexist Résumé Screener 99\u003c\/p\u003e \u003cp\u003eThe Necessity of Model Interpretability 101\u003c\/p\u003e \u003cp\u003eConnections Between Predictive Performance and Interpretability 103\u003c\/p\u003e \u003cp\u003eUniting (High) Model Performance and Model Interpretability 105\u003c\/p\u003e \u003cp\u003eCategories of Interpretability Methods 107\u003c\/p\u003e \u003cp\u003eGlobal Methods 107\u003c\/p\u003e \u003cp\u003eLocal Methods 113\u003c\/p\u003e \u003cp\u003eReal-World Successes of Interpretability Methods 113\u003c\/p\u003e \u003cp\u003eFacilitating Debugging and Audit 114\u003c\/p\u003e \u003cp\u003eLeveraging the Improved Performance of Black-Box Models 116\u003c\/p\u003e \u003cp\u003eAcquiring New Knowledge 116\u003c\/p\u003e \u003cp\u003eAddressing Critiques of Interpretability Methods 117\u003c\/p\u003e \u003cp\u003eExplanations Generated by Interpretability Methods Are Not Robust 118\u003c\/p\u003e \u003cp\u003eExplanations Generated by Interpretability Methods Are Low Fidelity 120\u003c\/p\u003e \u003cp\u003eThe Forking Paths of Model Interpretability 121\u003c\/p\u003e \u003cp\u003eThe Four-Measure Baseline 122\u003c\/p\u003e \u003cp\u003eBuilding Our Own Credit Scoring Model 124\u003c\/p\u003e \u003cp\u003eUsing Train-Test Splits 125\u003c\/p\u003e \u003cp\u003eFeature Selection and Feature Engineering 125\u003c\/p\u003e \u003cp\u003eBaseline Models 127\u003c\/p\u003e \u003cp\u003eThe Importance of Making Your Code Work for Everyone 129\u003c\/p\u003e \u003cp\u003eExecution Variability 129\u003c\/p\u003e \u003cp\u003eAddressing Execution Variability with Functionalized Code 130\u003c\/p\u003e \u003cp\u003eStochastic Variability 130\u003c\/p\u003e \u003cp\u003eAddressing Stochastic Variability via Resampling 130\u003c\/p\u003e \u003cp\u003eSummary 133\u003c\/p\u003e \u003cp\u003e\u003cb\u003ePart III EDS in Practice 135\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003e\u003cb\u003eChapter 6 Beginning a Responsible Data Science Project 137\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003eHow the Responsible Data Science Framework Addresses the Common Cause 138\u003c\/p\u003e \u003cp\u003eDatasets Used 140\u003c\/p\u003e \u003cp\u003eRegression Datasets—Communities and Crime 140\u003c\/p\u003e \u003cp\u003eClassification Datasets—COMPAS 140\u003c\/p\u003e \u003cp\u003eCommon Elements Across Our Analyses 141\u003c\/p\u003e \u003cp\u003eProject Structure and Documentation 141\u003c\/p\u003e \u003cp\u003eProject Structure for the Responsible Data\u003c\/p\u003e \u003cp\u003eScience Framework: Everything in Its Place 142\u003c\/p\u003e \u003cp\u003eDocumentation: The Responsible Thing to Do 145\u003c\/p\u003e \u003cp\u003eBeginning a Responsible Data Science Project 151\u003c\/p\u003e \u003cp\u003eCommunities and Crime (Regression) 151\u003c\/p\u003e \u003cp\u003eJustification 151\u003c\/p\u003e \u003cp\u003eCompilation 154\u003c\/p\u003e \u003cp\u003eIdentifying Protected Classes 157\u003c\/p\u003e \u003cp\u003ePreparation—Data Splitting and Feature Engineering 159\u003c\/p\u003e \u003cp\u003eDatasheets 161\u003c\/p\u003e \u003cp\u003eCOMPAS (Classification) 164\u003c\/p\u003e \u003cp\u003eJustification 164\u003c\/p\u003e \u003cp\u003eCompilation 166\u003c\/p\u003e \u003cp\u003eIdentifying Protected Classes 168\u003c\/p\u003e \u003cp\u003ePreparation 169\u003c\/p\u003e \u003cp\u003eSummary 172\u003c\/p\u003e \u003cp\u003e\u003cb\u003eChapter 7 Auditing a Responsible Data Science Project 173\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003eFairness and Data Science in Practice 175\u003c\/p\u003e \u003cp\u003eThe Many Different Conceptions of Fairness 175\u003c\/p\u003e \u003cp\u003eDifferent Forms of Fairness Are Trade-Offs with Each Other 177\u003c\/p\u003e \u003cp\u003eQuantifying Predictive Fairness Within a Data Science Project 179\u003c\/p\u003e \u003cp\u003eMitigating Bias to Improve Fairness 185\u003c\/p\u003e \u003cp\u003ePreprocessing 185\u003c\/p\u003e \u003cp\u003eIn-processing 186\u003c\/p\u003e \u003cp\u003ePostprocessing 186\u003c\/p\u003e \u003cp\u003eClassification Example: COMPAS 187\u003c\/p\u003e \u003cp\u003ePrework: Code Practices, Modeling, and Auditing 187\u003c\/p\u003e \u003cp\u003eJustification, Compilation, and Preparation Review 189\u003c\/p\u003e \u003cp\u003eModeling 191\u003c\/p\u003e \u003cp\u003eAuditing 200\u003c\/p\u003e \u003cp\u003ePer-Group Metrics: Overall 200\u003c\/p\u003e \u003cp\u003ePer-Group Metrics: Error 202\u003c\/p\u003e \u003cp\u003eFairness Metrics 204\u003c\/p\u003e \u003cp\u003eInterpreting Our Models: Why Are They Unfair? 207\u003c\/p\u003e \u003cp\u003eAnalysis for Different Groups 209\u003c\/p\u003e \u003cp\u003eBias Mitigation 214\u003c\/p\u003e \u003cp\u003ePreprocessing: Oversampling 214\u003c\/p\u003e \u003cp\u003ePostprocessing: Optimizing Thresholds\u003c\/p\u003e \u003cp\u003eAutomatically 218\u003c\/p\u003e \u003cp\u003ePostprocessing: Optimizing Thresholds Manually 219\u003c\/p\u003e \u003cp\u003eSummary 223\u003c\/p\u003e \u003cp\u003e\u003cb\u003eChapter 8 Auditing for Neural Networks 225\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003eWhy Neural Networks Merit Their Own Chapter 227\u003c\/p\u003e \u003cp\u003eNeural Networks Vary Greatly in Structure 227\u003c\/p\u003e \u003cp\u003eNeural Networks Treat Features Differently 229\u003c\/p\u003e \u003cp\u003eNeural Networks Repeat Themselves 231\u003c\/p\u003e \u003cp\u003eA More Impenetrable Black Box 232\u003c\/p\u003e \u003cp\u003eBaseline Methods 233\u003c\/p\u003e \u003cp\u003eRepresentation Methods 233\u003c\/p\u003e \u003cp\u003eDistillation Methods 234\u003c\/p\u003e \u003cp\u003eIntrinsic Methods 235\u003c\/p\u003e \u003cp\u003eBeginning a Responsible Neural Network Project 236\u003c\/p\u003e \u003cp\u003eJustification 236\u003c\/p\u003e \u003cp\u003eMoving Forward 239\u003c\/p\u003e \u003cp\u003eCompilation 239\u003c\/p\u003e \u003cp\u003eTracking Experiments 241\u003c\/p\u003e \u003cp\u003ePreparation 244\u003c\/p\u003e \u003cp\u003eModeling 245\u003c\/p\u003e \u003cp\u003eAuditing 247\u003c\/p\u003e \u003cp\u003ePer-Group Metrics: Overall 247\u003c\/p\u003e \u003cp\u003ePer-Group Metrics: Unusual Definitions of “False Positive” 248\u003c\/p\u003e \u003cp\u003eFairness Metrics 249\u003c\/p\u003e \u003cp\u003eInterpreting Our Models: Why Are They Unfair? 252\u003c\/p\u003e \u003cp\u003eBias Mitigation 253\u003c\/p\u003e \u003cp\u003eWrap-Up 255\u003c\/p\u003e \u003cp\u003eAuditing Neural Networks for Natural Language Processing 258\u003c\/p\u003e \u003cp\u003eIdentifying and Addressing Sources of Bias in NLP 258\u003c\/p\u003e \u003cp\u003eThe Real World 259\u003c\/p\u003e \u003cp\u003eData 260\u003c\/p\u003e \u003cp\u003eModels 261\u003c\/p\u003e \u003cp\u003eModel Interpretability 262\u003c\/p\u003e \u003cp\u003eSummary 262\u003c\/p\u003e \u003cp\u003e\u003cb\u003eChapter 9 Conclusion 265\u003c\/b\u003e\u003c\/p\u003e \u003cp\u003eHow Can We Do Better? 267\u003c\/p\u003e \u003cp\u003eThe Responsible Data Science Framework 267\u003c\/p\u003e \u003cp\u003eDoing Better As Managers 269\u003c\/p\u003e \u003cp\u003eDoing Better As Practitioners 270\u003c\/p\u003e \u003cp\u003eA Better Future If We Can Keep It 271\u003c\/p\u003e \u003cp\u003eIndex 273\u003c\/p\u003e\u003c\/font\u003e\u003c\/p\u003e\r\n\r\n\u003cp\u003e\u003cfont size=\"3\"\u003eSubject Areas: Business applications [\u003ca title=\"See our other books on Business applications\" href=\"https:\/\/freshlyprintedbooks.co.uk\/search?q=%22Business%20applications%20%5BUF%5D%22\"\u003eUF\u003c\/a\u003e]\u003c\/font\u003e\u003c\/p\u003e\r\n\r\n\r\n\u003c\/font\u003e","brand":"Wiley","offers":[{"title":"Brand New","offer_id":52174420607256,"sku":"9781119741756","price":23.29,"currency_code":"GBP","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0730\/2037\/5320\/files\/9781119741756.jpg?v=1781174070","url":"https:\/\/freshlyprintedbooks.co.uk\/products\/responsible-data-science-paperback-softback-9781119741756","provider":"Freshly Printed Books","version":"1.0","type":"link"}