Ralph Lauren
Market Basket Analysis

Data Scientist | Data Analysis | 2021
Client: Ralph Lauren
Skills Used: Python, AWS SageMaker, AWS S3

Overview

For the capstone project of my data science master's program at Columbia, I, along with a team of other master's students, worked as data science freelancers for Ralph Lauren. We analyzed and evaluated four tests performed in select Ralph Lauren stores using Amazon SageMaker and real customer data provided by Ralph Lauren's data analytics team. We were also built a general and scalable analysis framework that could be used to clean customer datasets and perform market basket analysis for future tests

The PRoblem

The four in-store tests are shown below:

I primarily worked on the footwear assortment expansion test, which involved examining changes in customer purchasing behavior when footwear was introduced into a non-footwear store.

The Solution

In footwear assortment expansion test, there were 5 types of footwear display types:

Boxes on the floor
Boxes off the floor
From our home to yours (FOHTY)
Try on ship to home
Full service door

The rollout period was during Q2 (April - June) and Q3 (July - September).

In our analysis, we found that few of all the transactions (1.5%) done in the rollout period (Q2-Q3) included an item of footwear.

From those transactions, sandals were the most popular footwear to buy (84% of footwear transactions). The figure below shows that for every display type, the most popular items sold were men’s and women’s sandals. Athletic shoes were below sandals in terms of popularity and casual shoes were below athletic shoes. We also noticed that most footwear baskets contained only one pair of footwear (82% of footwear transactions). Socks, dresses, underwear, caps, short-sleeve mesh t-shirts, and women's t-shirts were also found to drive footwear sales.

Sandals were slightly more likely to be bought from stores with boxes on the floor display or FOHTY display, and outdoor stores near beaches, rural areas, and metro areas. Athletic shoes were more likely to be sold from stores with a full service door display or boxes off the floor display and indoor domestic/international stores. In these figures purchase likelihood represents the chance a sold footwear item was purchased from that store type.

In conclusion, we found that sandals should be introduced into all stores, especially at outdoor, rural, beach stores with boxes on floor and FOTHY display. We also recommended to Ralph Lauren to introduce athletic shoes into international, domestic, and indoor stores with full service door and boxes off floor display and to include other types of footwear in select stores, since they were not very popular.