Statistics Slam Dunk: Statistical analysis with R on real NBA data

Statistics Slam Dunk....1

brief contents....8

contents....10

foreword....17

preface....19

acknowledgments....21

about this book....23

Who should read this book....23

How this book is organized: A road map....24

About the code....27

liveBook discussion forum....27

about the author....28

about the cover illustration....29

Chapter 1: Getting started....31

1.1 Brief introductions to R and RStudio....32

1.2 Why R?....34

1.2.1 Visualizing data....34

1.2.2 Installing and using packages to extend R’s functional footprint....37

1.2.3 Networking with other users....38

1.2.4 Interacting with big data....39

1.2.5 Landing a job....39

1.3 How this book works....39

Chapter 2: Exploring data....44

2.1 Loading packages....45

2.2 Importing data....46

2.3 Wrangling data....47

2.3.1 Removing variables....48

2.3.2 Removing observations....48

2.3.3 Viewing data....49

2.3.4 Converting variable types....50

2.3.5 Creating derived variables....51

2.4 Variable breakdown....53

2.5 Exploratory data analysis....55

2.5.1 Computing basic statistics....55

2.5.2 Returning data....58

2.5.3 Computing and visualizing frequency distributions....59

2.5.4 Computing and visualizing correlations....72

2.5.5 Computing and visualizing means and medians....75

2.6 Writing data....81

Chapter 3: Segmentation analysis....83

3.1 More on tanking and the draft....84

3.2 Loading packages....85

3.3 Importing and viewing data....86

3.4 Creating another derived variable....87

3.5 Visualizing means and medians....88

3.5.1 Regular season games played....88

3.5.2 Minutes played per game....92

3.5.3 Career win shares....94

3.5.4 Win shares every 48 minutes....96

3.6 Preliminary conclusions....98

3.7 Sankey diagram....99

3.8 Expected value analysis....104

3.9 Hierarchical clustering....110

Chapter 4: Constrained optimization....116

4.1 What is constrained optimization?....117

4.2 Loading packages....118

4.3 Importing data....119

4.4 Knowing the data....119

4.5 Visualizing the data....122

4.5.1 Density plots....122

4.5.2 Boxplots....125

4.5.3 Correlation plot....127

4.5.4 Bar chart....129

4.6 Constrained optimization setup....132

4.7 Constrained optimization construction....134

4.8 Results....139

Chapter 5: Regression models....142

5.1 Loading packages....144

5.2 Importing data....144

5.3 Knowing the data....145

5.4 Identifying outliers....148

5.4.1 Prototype....148

5.4.2 Identifying other outliers....152

5.5 Checking for normality....157

5.5.1 Prototype....158

5.5.2 Checking other distributions for normality....159

5.6 Visualizing and testing correlations....163

5.6.1 Prototype....164

5.6.2 Visualizing and testing other correlations....165

5.7 Multiple linear regression....167

5.7.1 Subsetting data into train and test....167

5.7.2 Fitting the model....168

5.7.3 Returning and interpreting the results....168

5.7.4 Checking for multicollinearity....171

5.7.5 Running and interpreting model diagnostics....172

5.7.6 Comparing models....173

5.7.7 Predicting....176

5.8 Regression tree....180

Chapter 6: More wrangling and visualizing data....184

6.1 Loading packages....185

6.2 Importing data....185

6.3 Wrangling data....186

6.3.1 Subsetting data sets....186

6.3.2 Joining data sets....187

6.4 Analysis....191

6.4.1 First quarter....191

6.4.2 Second quarter....195

6.4.3 Third quarter....198

6.4.4 Fourth quarter....199

6.4.5 Comparing best and worst teams....201

6.4.6 Second-half results....211

Chapter 7: T-testing and effect size testing....217

7.1 Loading packages....218

7.2 Importing data....219

7.3 Wrangling data....219

7.4 Analysis on 2018–19 data....221

7.4.1 2018–19 regular season analysis....221

7.4.2 2019 postseason analysis....227

7.4.3 Effect size testing....231

7.5 Analysis on 2019–20 data....233

7.5.1 2019–20 regular season analysis (pre-COVID)....233

7.5.2 2019–20 regular season analysis (post-COVID)....238

7.5.3 More effect size testing....241

Chapter 8: Optimal stopping....244

8.1 Loading packages....245

8.2 Importing images....245

8.3 Importing and viewing data....246

8.4 Exploring and wrangling data....247

8.5 Analysis....252

8.5.1 Milwaukee Bucks....252

8.5.2 Atlanta Hawks....259

8.5.3 Charlotte Hornets....263

8.5.4 NBA....267

Chapter 9: Chi-square testing and more effect size testing....271

9.1 Loading packages....272

9.2 Importing data....273

9.3 Wrangling data....274

9.4 Computing permutations....277

9.5 Visualizing results....279

9.5.1 Creating a data source....280

9.5.2 Visualizing the results....281

9.5.3 Conclusions....283

9.6 Statistical test of significance....285

9.6.1 Creating a contingency table and a balloon plot....286

9.6.2 Running a chi-square test....288

9.6.3 Creating a mosaic plot....288

9.7 Effect size testing....289

Chapter 10: Doing more with ggplot2....291

10.1 Loading packages....292

10.2 Importing and viewing data....292

10.3 Salaries and salary cap analysis....294

10.4 Analysis....300

10.4.1 Plotting and computing correlations between team payrolls and regular season wins....301

10.4.2 Payrolls versus end-of-season results....312

10.4.3 Payroll comparisons....315

Chapter 11: K-means clustering....323

11.1 Loading packages....324

11.2 Importing data....324

11.3 A primer on standard deviations and z-scores....325

11.4 Analysis....327

11.4.1 Wrangling data....328

11.4.2 Evaluating payrolls and wins....332

11.5 K-means clustering....337

11.5.1 More data wrangling....339

11.5.2 K-means clustering....341

Chapter 12: Computing and plotting inequality....352

12.1 Gini coefficients and Lorenz curves....353

12.2 Loading packages....354

12.3 Importing and viewing data....355

12.4 Wrangling data....357

12.5 Gini coefficients....363

12.6 Lorenz curves....367

12.7 Salary inequality and championships....371

12.7.1 Wrangling data....372

12.7.2 T-test....376

12.7.3 Effect size testing....379

12.8 Salary inequality and wins and losses....380

12.8.1 T-test....380

12.8.2 Effect size testing....382

12.9 Gini coefficient bands versus winning percentage....383

Chapter 13: More with Gini coefficients and Lorenz curves....388

13.1 Loading packages....389

13.2 Importing and viewing data....390

13.3 Wrangling data....390

13.4 Gini coefficients....396

13.5 Lorenz curves....399

13.6 For loops....403

13.6.1 Simple demonstration....404

13.6.2 Applying what we’ve learned....404

13.7 User-defined functions....408

13.8 Win share inequality and championships....411

13.8.1 Wrangling data....412

13.8.2 T-test....417

13.8.3 Effect size testing....420

13.9 Win share inequality and wins and losses....423

13.9.1 T-test....423

13.9.2 Effect size testing....426

13.10 Gini coefficient bands versus winning percentage....426

Chapter 14: Intermediate and advanced modeling....431

14.1 Loading packages....432

14.2 Importing and wrangling data....432

14.2.1 Subsetting and reshaping our data....433

14.2.2 Extracting a substring to create a new variable....435

14.2.3 Joining data....436

14.2.4 Importing and wrangling additional data sets....436

14.2.5 Joining data (one more time)....439

14.2.6 Creating standardized variables....440

14.3 Exploring data....441

14.4 Correlations....445

14.4.1 Computing and plotting correlation coefficients....445

14.4.2 Running correlation tests....449

14.5 Analysis of variance models....450

14.5.1 Data wrangling and data visualization....451

14.5.2 One-way ANOVAs....454

14.6 Logistic regressions....458

14.6.1 Data wrangling....459

14.6.2 Model development....460

14.7 Paired data before and after....471

Chapter 15: The Lindy effect....477

15.1 Loading packages....479

15.2 Importing and viewing data....479

15.3 Visualizing data....482

15.3.1 Creating and evaluating violin plots....483

15.3.2 Creating paired histograms....484

15.3.3 Printing our plots....485

15.4 Pareto charts....487

15.4.1 ggplot2 and ggQC packages....488

15.4.2 qcc package....490

Chapter 16: Randomness versus causality....494

16.1 Loading packages....495

16.2 Importing and wrangling data....496

16.3 Rule of succession and the hot hand....498

16.4 Player-level analysis....503

16.4.1 Player 1 of 3: Giannis Antetokounmpo....503

16.4.2 Player 2 of 3: Julius Randle....508

16.4.3 Player 3 of 3: James Harden....512

16.5 League-wide analysis....515

Chapter 17: Collective intelligence....520

17.1 Loading packages....521

17.2 Importing data....522

17.3 Wrangling data....522

17.4 Automated exploratory data analysis....526

17.4.1 Baseline EDA with tableone....526

17.4.2 Over/under EDA with DataExplorer....529

17.4.3 Point spread EDA with SmartEDA....543

17.5 Results....552

17.5.1 Over/under....553

17.5.2 Point spreads....564

Chapter 18: Statistical dispersion methods....570

18.1 Loading a package....571

18.2 Importing data....571

18.3 Exploring and wrangling data....572

18.4 Measures of statistical dispersion and intra-season parity....576

18.4.1 Variance method....576

18.4.2 Standard deviation method....579

18.4.3 Range method....581

18.4.4 Mean absolute deviation method....583

18.4.5 Median absolute deviation method....586

18.5 Churn and inter-season parity....588

18.5.1 Data wrangling....589

18.5.2 Computing and visualizing churn....590

Chapter 19: Data standardization....595

19.1 Loading a package....596

19.2 Importing and viewing data....597

19.3 Wrangling data....598

19.3.1 Treating duplicate records....598

19.3.2 Final trimmings....602

19.4 Standardizing data....603

19.4.1 Z-score method....605

19.4.2 Standard deviation method....608

19.4.3 Centering method....610

19.4.4 Range method....612

Chapter 20: Finishing up....616

20.1 Cluster analysis....617

20.2 Significance testing....619

20.3 Effect size testing....622

20.4 Modeling....624

20.5 Operations research....627

20.6 Probability....630

20.7 Statistical dispersion....632

20.8 Standardization....633

20.9 Summary statistics and visualization....635

Appendix: More ggplot2 visualizations....639

index....661

Numerics....661

A....661

B....661

C....661

D....662

E....664

F....664

G....664

H....665

I....665

J....665

K....665

L....666

M....666

N....667

O....667

P....667

Q....668

R....668

S....668

T....669

U....670

V....670

W....670

X....670

Y....670

Z....670

Statistics Slam Dunk is a data science manual with a difference. Each chapter is a complete, self-contained statistics or data science project for you to work through—from importing data, to wrangling it, testing it, visualizing it, and modeling it. Throughout the book, you’ll work exclusively with NBA data sets and the R language, applying best-in-class statistics techniques to reveal fun and fascinating truths about the NBA.

About the book

Is losing basketball games on purpose a rational strategy? Which hustle statistics have an impact on wins and losses? Does spending more on player salaries translate into a winning record? You’ll answer all these questions and more. Plus, R’s visualization capabilities shine through in the book’s 300 plots and charts, including Pareto charts, Sankey diagrams, Cleveland dot plots, and dendrograms.

What's inside

• Transforming, tidying, and wrangling data
• Applying best-in-class exploratory data analysis techniques
• Developing supervised and unsupervised machine learning algorithms
• Executing hypothesis tests and effect size tests

About the reader

For readers who know basic statistics. No advanced knowledge of R—or basketball—required.