Scholarlink

<p>Customer churn prediction is a critical issue for businesses in terms of increasing customer loyalty and reducing revenue loss. Being able to predict the likelihood of a customer leaving a service allows businesses to develop important strategies and increase customer satisfaction. Minimising customer churn not only reduces costs but also builds a strong customer base for long-term business success. In this study, the performance of various machine learning algorithms in predicting customer churn is compared. Logistic Regression, Random Forest, Support Vector Machines (SVM), XGBoost, LightGBM, CatBoost, AdaBoost and K-Nearest Neighbour (KNN) algorithms were used in the study and performance metrics such as accuracy, precision, recall, F1 score and ROC AUC score were calculated and compared for each model. The findings show that XGBoost (94%), LightGBM (95%) and CatBoost (94%) algorithms perform better than other algorithms in predicting customer churn. The algorithms used in this context can help businesses to determine the most appropriate models to prevent customer churn. In addition, the use of such models can strengthen long-term customer relationships by increasing customer satisfaction.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Forecasting customer churn using machine learning: a comparative modeling approach

  • 期刊论文

  • 发表日期：

    2026年8月3日

查看全文
  • Murat Kilinc

摘要

Customer churn prediction is a critical issue for businesses in terms of increasing customer loyalty and reducing revenue loss. Being able to predict the likelihood of a customer leaving a service allows businesses to develop important strategies and increase customer satisfaction. Minimising customer churn not only reduces costs but also builds a strong customer base for long-term business success. In this study, the performance of various machine learning algorithms in predicting customer churn is compared. Logistic Regression, Random Forest, Support Vector Machines (SVM), XGBoost, LightGBM, CatBoost, AdaBoost and K-Nearest Neighbour (KNN) algorithms were used in the study and performance metrics such as accuracy, precision, recall, F1 score and ROC AUC score were calculated and compared for each model. The findings show that XGBoost (94%), LightGBM (95%) and CatBoost (94%) algorithms perform better than other algorithms in predicting customer churn. The algorithms used in this context can help businesses to determine the most appropriate models to prevent customer churn. In addition, the use of such models can strengthen long-term customer relationships by increasing customer satisfaction.