​

Analyzing the Shopping Habits of IIMB Students at campus MARS store

GROUP 4

Motivation & Objectives

• Classification of customers on basis of buying behaviour and other attributes
• Difference in buying actions on basis of academic examination, seasonality, balance availability etc.

​

​

Our goal:

improve sales performance at Mars and �optimize stocking

Data Set & Clean-up

Raw data:

• 55023 transactions
• >1000 different products
• Purchasing date, customer id, product name and id, quanity and price per sold product

​

Cleaned data set:

• 405 customers
• 7 product categories
• Amount per purchase
• Purchase time
• Purchase frequency
• Time variables
• Category sales and ratios

​

​

Current Trends at Mars store

Special events have an impact on the daily sales

In the Clothing and SOI category are mainly high-value products

Daily Sales in the year

Discriminant analysis

• The predictive power of the model is not up to the mark
• but is better than random selection when it comes to classifying females impacting the change

Regression analysis

1. Regression: Daily footfall and time variable

2. Regression: Frequency of visit and customer spending ratios

• The model has reasonable explanation power
• Food and personal care item variety drives up customer visits

Cluster Analysis

• Strive to bring the customers from low to a higher cart-size
• Promote clothing wear among low spenders to convert to mid category
• high spenders cannibalize Pcare spending from clothing products

Canonical Correlation

​

​

​

​

​

• Term start and term end most important variables
• Stocking to be increased at term beginning and end

Recommendations

• More clothing deals and marketing to targeted audience
• Increasing spending limit for high spending groups
• Term Start and Term end
• Increase variety of food items and personal care products

​

​