Understanding Consumer Behavior in the Retail Sector Using RFM Segmentation and Machine Learning: An Analysis

Abstract

The current study uses advanced analytics to examine the intricate dynamics of consumer behavior in the Indian retail sector. We created a preprocessed retail dataset with 495,478 retail clients in it.  The study aims to forecast consumer behavior, the study uses a variety of techniques, such as temporal analysis, Box-Cox transformation, and a marketing analysis method called RFM (Recency, Frequency, Monetary) segmentation. Additionally, it makes use of a variety of supervised machine learning models, such as RandomForest, AdaBoost, ExtraTrees, LGBM, and XGBoost, among which ET and XGBClassifier have demonstrated the highest levels of accuracy in customer lifetime value predicting. The study stated that the machine learning models' performance measures are remarkable: 92.40% accuracy, 92.27% precision, 92.40% recall, 92.28% F1 score, and 97.39 AUC. The study's findings validated the durability of machine learning techniques and demonstrated the model's accuracy in predicting customer lifetime value clusters. Important conclusions from RFM analysis show that it has a special value in offering important insights into customers and their behavior. This study sets a new benchmark for retail analytics by providing a scalable and effective technique for research projects in the future that use data analytics to understand and forecast customer behavior across various business entities.