E-commerce platform

Overview

An upcoming local brand company wanted to predict the products most likely to be purchased by their users to manage their inventory at the store. At the same time, the company chose to implement recommendation engine like Amazon, Target etc. The company would use the insights generated to cluster the data based on their history of purchase and target recommendations based on the outcome of statistical models.

Challenge

Since it’s a local brand company who has the potential to become a big name in the future, they could not afford to outsource or hire an analytics team. The company is running for 6 years and have collected the data in more than 4 systems due to the changing management practices. Tracking back with the versioning system is another major challenge.

Implementation

The team of two were hired- a data analyst and a data scientist who complimented each other’s work. The analyst blended the data from multiple sources and produced it in a format the data scientist had to work in. The data scientist then extracted various data points like the customer data-

• What are the previous purchases done by the customer?
• Does the customer have a pattern in shopping items?
• Does the customer purchase during the promotions?

A model was reused using R libraries and implemented which helped the manager answer the questions. Every time the process flow refreshed; the recommendation was sent to the user in their email ID. The process flow refresh would happen only when the last login activity of the user was a day before. The entire process implementation would leave the client with a huge potential on implementing the recommendation model into the platform (and not just an email), once they were ready.

Benefits Achieved

• Automated process
• Recommendation set up using R
• Boosting the sales by 8%