Identification of an Efficient Machine Learning Algorithm for the Prediction of Customer Churn: A Case Analysis of a Business-to-Consumer (B2C) Dairy Company for Customer Retention Strategy
S. K. Kamalakhannan,V. Rajkiran,A. M. Faisal,P. Elantheraiyan,K. S. Singh
DOI: https://doi.org/10.1109/ICIPTM59628.2024.10563757
2024-02-21
Abstract:Machine learning algorithms can be applied to real-life applications, such as customer churn rates for dairy products, which are part of the business-to-consumer (B2C) model, to create a strong customer retention strategy. The ultimate objective is to identify an efficient machine learning algorithm that may be utilized for predicting consumer churn in dairy-based products. Following the exploratory data analysis (EDA) phase, churn prediction models are constructed using machine learning algorithms such as logistic regression, random forest, support vector machine, K-nearest neighbor (KNN), decision tree classifier, and Gaussian naive bayes. The Kaplan-Meier curve, which is based on the customer lifetime value (CLV), is used to understand the churn behavior of the customers. Two binary classifier models are chosen for further study: random forest and logistic regression, determined by the area under the receiver operating characteristic curve (AUROC). Comparing the random forest to the logistic regression, the random forest performs better overall (87.26%), with precision (85.44%), recall (88%), and F1 score (86.7%). The random forest classifier has an AUC value of 95%, which is higher than the AUC score of 93% for the refined logistic regression model, as seen by the ROC graph. In comparison to the logistic regression, the random forest classifier is doing rather well on this binary class classification task based on the confusion matrix and ROC graph. Thus, the random forest classifier is considered to be the most effective one for dairy products.
Computer Science,Business