Cross-Selling Strategies in e-Marketing 1. Market Basket Analysis  Correlation Matrix  Factor Analysis  Conjoint Analysis 2. Profiling and Cloning Customers  Automatic Cluster Detection 2019/2/25

Profiling and Cloning Customers Factor Analysis vs. Cluster Analysis Measures of Similarity Clustering Based on Observed Data Clustering Based on Estimated Parameters Clustering Based on Observations and Estimates Simultaneously Methods for Classification 2019/2/25

Factor Analysis vs. Cluster Analysis X1 X2 X3 X4 X5 X6 X7 F1 F2 F3 F4 1 1 2 2 3 3 4 4 Customer 5 5 Factor Analysis i i N N 2019/2/25

Factor Analysis vs. Cluster Analysis X1 X2 X3 X4 X5 X6 X7 X1 X2 X3 X4 X5 X6 X7 1 2 G1 3 4 Customer 5 G2 Cluster Analysis i G3 G4 N 2019/2/25

Cluster Analysis Cluster analysis is a technique for grouping individuals or subjects into unknown groups. 2019/2/25

Similarity Measures X2 (X11, X12) (X21, X22) X1 2019/2/25

Similarity Measures In general, the distance measure between the ith and jth customer under p-dimension structure is 2019/2/25

Structural Similarity Ward’s Method Minimize within cluster variation 2019/2/25

Clustering Based on Observed Data 2019/2/25

2019/2/25

2019/2/25

Hierarchical Cluster Analysis of Observations Euclidean Distance, Ward Linkage Amalgamation Steps 2019/2/25

2019/2/25

連鎖咖啡店消費行為研究 距離上次到「連鎖咖啡店」已相隔多久時間？ 在過去6個月內到「連鎖咖啡店」消費的次數？ 過去到「連鎖咖啡店」消費，平均每次約多少錢？ 每次在咖啡店內約停留多久時間？ 每次到咖啡店，同行人數約為多少人? 2019/2/25

距離上次到「連鎖咖啡店」已相隔多久時間？ 2019/2/25

在過去6個月內到「連鎖咖啡店」消費的次數？ 2019/2/25

過去到「連鎖咖啡店」消費，平均每次約多少錢？ 2019/2/25

每次在咖啡店內約停留多久時間？ 2019/2/25

每次到咖啡店，同行人數約為多少人? 2019/2/25

Profiling and Cloning Customers Final Cluster Centers Cluster 1 2 3 4 7.56 4.15 2.40 2.44 2.05 3.67 4.40 5.44 2.85 2.85 2.30 2.44 3.95 3.26 4.30 4.22 3.05 2.85 3.20 1.11 距離上次到咖啡店相隔多久時間？ 在過去6個月內到咖啡店消費的次數？ 平均每次到咖啡店消費約多少錢？ 每次在咖啡店內約停留多久時間？ 每次到咖啡店，同行人數約為多少人? 總市場比率： 品牌A比率： 品牌B比率： 25% 25% 25% 25% 10% 20% 30% 40% 40% 30% 20% 10% 2019/2/25

Clustering Based on Estimated Parameters Clustering based on b1, b2, …, bp 2019/2/25

Cluster Regression Model Clustering Based on Observations and Estimates Simultaneously Cluster Regression Model Group 1: Group 2: Group M: 2019/2/25

Profiling and Cloning Customers X1 X2 X3 X4 X5 X6 X7 X1 X2 X3 X4 X5 X6 X7 1 1 2 2 3 3 4 4 Customer 5 5 Cluster Analysis i i N N N+1 ? N+2 New Customers ? ? 2019/2/25 N+m ?

Methods for Classification Discriminant Analysis Multi-nominal Logit Regression Model 2019/2/25