Predicting Customer Churn in E-Commerce Using Machine Learning

Challenge Background

In Saudi Arabia, there are two platforms that provide customized online stores which are Zid and Salla. Those platforms have thousands of stores. We want to help small businesses with open-source machine learning models, especially those who don't have a data team. As we know the customer acquisition cost is higher than the customer retention cost. So, we decided that our first project will be about e-commerce customer churn prediction.‏

The Problem

In the domain of e-commerce, acquiring a new customer is generally more expensive than keeping the existing ones. The customers usually leave if they do not get good incentives. Thus, analyzing customer behavior to predict customer churn and the reasons can be a great solution for businesses especially small businesses and startups to monitor customer behavior and offer a suitable incentive that could help in maintaining the customers. In Saudi Arabia, most of the e-commerce platforms don’t offer an analytics tool for the traders to help them analyze the customer behavior which lead them to close their stores at the end. Therefore, offering a tool that can help them to analyze customer behavior will be a great contribution.

Goal of the Project

• Help small businesses in e-commerce.
• Reduce churn rates.
• Improve the skills of the team by sharing knowledge and overcoming challenges together.

Project Timeline

What you'll learn

1. Building an efficient machine learning pipeline.
2. Feature engineering.
3. Building machine learning model.