Time-aware Conversion Prediction For E-commerce

by Wendi Ji, Xiaoling Wang, Aoying Zhou

Overview

This unique compendium provides a novel research on how time influences the conversions of advertising and product recommendation in E-commerce. It proposes time-aware conversion prediction models to solve the problem — what products should be recommended for a given period to maximize conversion? The volume also presents a series of researches on how to build data-driven attribution models to allocate the time-sensitive contribution of advertisements to the conversion. This must-have reference text will be invaluable for researchers, professionals, academics and graduate students keen in databases and artificial intelligence.

Product Details

ISBN-13:	9789813224704
Publisher:	World Scientific Publishing Company, Incorporated
Publication date:	03/08/2018
Series:	East China Normal University Scientific Reports , #7
Pages:	152
Product dimensions:	6.10(w) x 9.10(h) x 0.60(d)

Table of Contents

Foreword v

About the Authors ix

1 Introduction 1

1.1 The Research Framework on Time-Aware Conversion Prediction 2

1.2 The Present State and Prospects 5

1.2.1 Computational Advertising 5

1.2.2 Product Recommendation 7

1.2.3 Attribution Analysis of Conversions 10

1.3 Techniques and Challenges in Time-Aware Conversion Prediction 12

1.3.1 Conversion Prediction in Computational Advertising 13

1.3.2 Conversion Prediction in Product Recommendation 13

1.3.3 Time-Aware Conversion Prediction 14

1.3.4 MTA Analysis Models of Conversions 16

1.4 The Structure of this Book 17

2 Basic Conversion Prediction Models 19

2.1 Framework of Computational Advertising and Recommendation 19

2.1.1 Paid Search Ads 21

2.1.2 Display Ads 22

2.1.3 Recommendation 25

2.2 Conversion Prediction Models 27

2.2.1 Feature Representation 27

2.2.2 Linear Models 29

2.2.3 Nonlinear Models 32

2.2.4 Conversion Rate Prediction and Click-Through Rate Prediction 36

2.3 Summary 38

3 Modeling of the Conversion Delay 39

3.1 Conversion Delay and Attribution Models 40

3.1.1 Modeling Conversion Delay 40

3.1.2 Attribution Analysis of Conversions 41

3.2 Survival and Event History Analysis 42

3.2.1 Hazard Rate and Survival Function 43

3.2.2 Parametric Lifetime Models 46

3.2.3 Temporal Point Processes 50

3.3 Summary 56

4 Time-Aware Conversion Prediction 57

4.1 User Behavior Sequence 59

4.2 Rank-Based Conversion Prediction Models 62

4.3 Mixture Temporal Mode 64

4.3.1 Conversion Intervals 65

4.3.2 Weibull Distribution 67

4.3.3 cWMM 69

4.4 Experiments 73

4.4.1 Datasets of Sequential Behaviors 73

4.4.2 Effectiveness of Lifetime Models 74

4.4.3 Just-in-Time Conversion Rate Prediction 79

4.4.4 Just-in-Time Product Recommendations 80

4.5 Summary 82

5 Multi-touch Attribution Analysis in Online Advertising 83

5.1 Data-Drive MTA 84

5.1.1 Modeling the Influence of Advertisements 84

5.2 Probabilistic MTA Model 86

5.2.1 MTA Model 86

5.2.2 Parameter Estimation 90

5.2.3 MTA 91

5.3 Experiments 91

5.3.1 Dataset 92

5.3.2 Baseline Methods 94

5.3.3 Choice of Probability Distribution 94

5.3.4 Interpretation of Model Parameters 95

5.3.5 Conversion Prediction 97

5.3.6 Attribution Analysis 101

5.4 Summary 103

6 Additional MTA Analysis in Online Advertising 105

6.1 AMTA Attribution 106

6.1.1 Survival Analysis 107

6.1.2 Additional Influence of Ad Exposures 108

6.1.3 Conversion Modeling Based on the AMAM 109

6.1.4 Parameter Estimation 111

6.2 Experiments 113

6.2.1 Dataset 113

6.2.2 Baseline Methods 114

6.2.3 Interpretation of Model Parameters 115

6.2.4 Conversion Rate Prediction 116

6.2.5 Attribution Analysis 117

6.3 Summary 118

7 Conclusions 119

References 123

Index 133