Introduction
-
Motivation
Retailers want to understand the customer purchase behavior (specifically, purchase amount) against various products of different categories. It's important for them building a model to predict the purchase amount of customer against various products which will help them to create personalized offer for customers against different products.
Our goal is to: 1. predict the customer purchase behavior when the customer demographics (age, gender, marital status, city_type), product details (product_id and product category) are given; 2. predict purchase solely based on the customer demographics so that they can reflect those in retailer's marketing strategies that target different consumers.
-
Solutions
-
Preprocess: We remap the attributes value from object(string) into integer with different techniques: label encoder for User_ID, Product_ID, Age ,Stay_in_Current_City_Years; one-hot-encoded for other nominal categorical attributes.
-
Using Collaborative Filtering to achieve the first goal, wrote by ourselves. Firstly, find the Users who bought the same or similar products, then calculate Cosine the similarity according to customer demographics and find three examples in training set. Finally, calculating the mean purchase of the three most similar User.
-
Using Random Forest to achieve the first goal with sklearn package.
-
Using SVM, Decision Tree and KNN with sklearn package to achieve the second goal, only using the low-level information.
-
You can find the code here​
-
You can download the full report by clicking here.