Customer Churn Reduction in Telecom

Abstract

In telecom industry, the key focus is to retain the customer at any cost. It's more difficult and expensive to attain new customers than retaining the old customers. So, the company aims to avoid the churn as much as possible. Retaining the customers and avoiding the churn needs the involvement of Data Mining techniques. To understand the customer behavior, we need the involvement of understanding and interpreting information from the data. So, we aim to understand the patterns and behavior of customers to avoid churn. In our project we gathered the data of telecom industry from Kaggle. Data attained needed Preprocessing. after the data is cleaned and ready for use, then applied Exploratory Data Analysis, then were able to discover the patterns that are leading to customer churn and from that we can bring customer retention. This exploratory data analysis let us understand the behavior of customers, using the telecom services. After that applied the machine learning models including Decision Tree, XGBoost and Random Forest on the data features to predict customer churn. Random forest and Decision Tree both outperformed XGBoost because of the best accuracy than XGBoost. After that implemented a web application using Flask web development framework aiming with the features that is convenient for a layman person for prediction understanding and data analysis. Keywords: Customer Churn in Telecom, Kaggle, Exploratory Data Analysis using Python, Random Forest, XGBoost, Decision Tree, Machine Learning, Flask Web development