Modeling techniques in predictive analytics business problems and solutions with R

Alpha 925/03-07-2014

Table of Contents
Preface v
Figures ix
Tables xiii
Exhibits xv
1. Analytics and Data Science 1
2. Advertising and Promotion 15
3. Preference and Choice 29
4. Market Basket Analysis 37
5. Economic Data Analysis 53
6. Operations Management 67
7. Text Analytics 83
8. Sentiment Analysis 113
9. Sports Analytics 149
10. Brand and Price 173
11. Spatial Data Analysis 209
12. The Big Little Data Game 231
A. There's a Pack for That 237
B. Measurement 253
C. Code and Utilities 267
Bibliography 297
Index 327

Today, successful firms win by understanding their data more deeply than competitors do. In short, they compete based on analytics. Now, in Modeling Techniques in Predictive Analytics, the leader of Northwestern University’s prestigious analytics program brings together all the concepts, techniques, and R code you need to excel in analytics. Thomas W. Miller’s unique balanced approach combines business context and quantitative tools, appealing to managers, analysts, programmers, and students alike.

Miller addresses multiple business challenges and business cases, including segmentation, brand positioning, product choice modeling, pricing research, finance, sports, Web and text analytics, and social network analysis. He illuminates the use of cross-sectional data, time series, spatial, and even spatio-temporal data. For each problem, Miller explains:
Why the problem is significant
What data is relevant
How to explore your data
How to model your data – first conceptually, with words and figures; and then with mathematics and programs
Miller walks through model construction, explanatory variable subset selection, and validation, demonstrating best practices for improving out-of-sample predictive performance. He employs data visualization and statistical graphics in exploring data, presenting models, and evaluating performance. Extensive example code is presented in R, today’s #1 system for applied statistics, statistical research, and predictive modeling; all code is set apart from other text so it’s easy to find for those who want it (and easy to skip for those who don’t).