Customer Retention Analysis Through Machine Learning-Based Churn Prediction

Many businesses are struggling with customer defection/turnover; this problem is widespread amongst competing organisations due to the currently increasing level of competition which adds an extra challenge for organisations attempting to deal with these problems. The purpose of this research paper is to examine how machine learning classification algorithms can be used to classify and predict the likelihood of customer turnover based on a variety of customer attributes. The initial step toward reaching the goals of this research paper was to find the best possible classification algorithm that could be used to predict service levels for customers in the future; thus providing information about which classification algorithm will give the most accurate predictions of classifications.
Several classification algorithms were reviewed in the course of this research paper and three classification algorithms were selected as the main focus of this research paper through further review (i.e., Logistic Regression (LR), Decision Tree (DT) and Random Forest (RF)), and to evaluate these algorithms at three different levels to compare the performance of the classification algorithms based upon common classification prediction measurement (i.e., accuracy, precision and total errors) to determine which classification algorithm produced the most accurate predicted classifications.