CHURN PREDICTION - A COMPARATIVE ANALYSIS WITH SUPERVISED MACHINE LEARNING ALGORITHMS

Customer churn predictive model plays an indispensable role in all the industries since "churn is the rate at which the customers stop doing business with an organization". Machine Learning algorithms are used to build faultless models for prediction and classification. In this paper, a comparative analyzation of the performance of five different supervised machine learning algorithms namely Gaussian Naive Bayes, Support Vector Machine, K Nearest Neighbours, Decision Tree and Random Forest Classifiers in predicting churn is studied. Churn_Modelling dataset from Kaggle is used to test these classifiers. Experimental outcomes show that Random Forest Classifier outperforms all other algorithms in predicting the churn of a customer regarding accuracy, precision and recall.