Customer churn analysis with data mining methods: Software as a service(SAAS) industry

Ensuring customer continuity and increasing the number of loyal customers is an essential issue for industries with increasing competitiveness and rapidly growing. Since the cost of acquiring new customers is much higher than retaining existing customers, businesses must examine existing customer behaviors, identify customers likely to leave the business, and conduct marketing activities to increase satisfaction. In the literature, some approaches are suggested to detect customers who have the intention to leave the company. The churn analysis method is one of them and is mostly used in business-to-consumer business models such as telecommunications, banking, and retailing. This study considered a software company that provides serviceswithin the business-to-business model was considered. Seventeen features in the data set with 1951 records were analyzed, and the ten features (number of products, number of customers, number of offers, number of orders, number of invoices, cargo usage, number of users, custom report usage, number of cash register receipts, email connection) that found a significant relationship with customer churn variable were selected to analyze in the study. Customer churn analysis was performed using Decision Tree, Random Forest, Logistic Regression, k-nearest neighbors (k-NN), Naive Bayes, and Artificial Neural Networks algorithms. As a result, the Random Forest algorithm was found to give the best result.