CUSTOMER SATISFACTION PREDICTION USING MACHINE LEARNING TECHNIQUES

Abstract

Businesses increasingly focus on understanding customer behavior and experience, especially factors contributing to stronger customer satisfaction. The prediction of satisfaction is regarded as a complex, but important task due to its critical role in shaping loyalty and retention by increasing service quality. Despite extensive research on consumer behavior, satisfaction remains a relatively underexplored target in the context of accurate prediction for commercial organizations. Data-driven approaches, specifically machine learning and deep learning strategies, have shown great potential in the analysis of customer feedback data, including techniques such as Logistic Regression, Decision Tree, Random Forest, XGBoost, SVM, ANN, LSTM, and etc. This study focuses on using machine learning methods to predict contentment levels by analyzing publicly available datasets, relevant to existing works. The initial four methods discussed earlier are applied and compared. The process involves selecting suitable datasets across various industries, preparing features, training the models, and evaluating their performance. Previously published works have been successfully replicated, with explanations offered for any discrepancies in outcomes.