Implementation of Data Mining for Churn Prediction in Music Streaming Company Using 2020 Dataset

Abstract

Customer is an important asset in a company as it is the lifeline of a company. For a company to get a new customer, it will cost a lot of money for campaigns. On the other hand, maintaining old customer tend to be cheaper than acquiring a new one. Because of that, it is important to be able to prevent the loss of customers from the products we have. Therefore, customer churn prediction is important in retaining customers. This paper discusses data mining techniques using XGBoost, Deep Neural Network, and Logistic Regression to compare the performance generated using data from a company that develops a song streaming application. The company suffers from the churn rate of the customer. Uninstall rate of the customers reaching 90% compared to the customer’s installs. The data will come from Google Analytics, a service from Google that will track the customer’s activity in the music streaming application. After finding out the method that will give the highest accuracy on the churn prediction, the attribute of data that most influence on the churn prediction will be determined.