Customer churn prediction in telecommunication industry using application of machine learning

Abstract

The concept of customer churn in telecommunications is important because it can help organizations better understand their customers' needs and preferences. It can also be used to develop more effective marketing strategies and improve customer service. Using customer churn prediction models can also help companies reduce the cost of customer acquisition and retention. This paper describes how to build a data analysis model that predicts early subscribers leaving a telecommunications network using different machine learning algorithms commonly used such as Random Forest, Support Vector Machine, Artificial Neural Network, Logistics Regression, and K-Nearest Neighbor. The dataset was also subjected to resampling techniques for dealing with imbalanced data, such as Synthetic Minority Oversampling Technique Tomek Link (SMOTE Tomek), Synthetic Minority Oversampling Technique ENN (SMOTE ENN) and Adaptive Synthetic Sampling Approach for Imbalanced Learning (ADASYN). Finally, the test set results have been evaluated using a confusion matrix and an AUC curve.