Improve customer churn prediction through the proposed PCA-PSO-K means algorithm in the communication industry

Customer churn prediction is one of the areas in Customer Relationship Management that differentiates loyal customers from factors that have a negative impact on business growth. Hence, various machine learning-based methods have been developed by researchers to accurately predict customer churn. However, high dimensionality and low prediction accuracy are problems in identifying averse customers. This paper presents a new system called PCA-PSO-K Means algorithm, which combines three algorithms: principal component analysis (PCA) for data set feature reduction, K Means algorithm for classification, and particle swarm optimization (PSO) algorithm to optimize K Means in providing initial centroids. The experimental results in the data set of one of the fixed internet providers in Isfahan Province show the improvement of the accuracy of customer churn prediction. The proposed system has an accuracy of 99.77%, a sensitivity of 75%, a specificity of 99.81% and a correlation coefficient of 0.443 +/- 0.271. Found.