A grid search optimized extreme learning machine approach for customer churn prediction

Behavioral traits of customers, such as loyalty status and satisfaction criteria are subject to alterations due to the rapidly changing world. Therefore, these behavioral changes should be analyzed efficiently at every step of the decision-making process. Customer churn analysis involves the determination of customers who tend to leave a situation before it occurs, by analyzing customer data using various methods. The aim of this study is to develop an extreme learning machine-based model to analyze the customer churn prediction problem and determine the parameters that improve the performance of the model. Grid search is used for hyperparameter tuning. In addition, a modified accuracy calculation approach is presented. In this study, we have developed various models based on Naïve Bayes, k-nearest neighbor, and support vector machine methods and provided a comparison of the performance of each model. According to the results obtained, an accuracy of 93.1% was achieved using the proposed extreme learning machine model.”. In addition, the proposed model is highly effective in solving the research problem because, the number of parameters to be determined is less, thus reducing its competition with other models.