University of Pittsburgh Cs2750 Machine Learning Handout 15 Professor Milos Hauskrecht Problem Assignment 8 Problem 1. Customer Profiling and Predictions with a Latent Variable Model

In this problem we use the the dataset of consumers preferences for different product brands. The dataset consists of 6 attributes corresponding to individual product categories, the values correspond to different brands in respective categories. For the sake of simplicity, we use numerical labels to distinguish the brands. We assume that each product can appear only in one category. Each customer corresponds to one entry (row) in the dataset. The number of brand products in each product category is listed in the following table.