Unlocking Retail Insights: Predictive Modeling and Customer Segmentation Through Data Analytics
This research aims at examining the progress of retail demand forecasting and customer classification via regression models and RFM analysis in the retail chain industry. Entailing actual retail sales data, this work utilizes three regression models:—MLP Regressor, Ridge Regressor, and KNN Regressor...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-03-01
|
| Series: | Journal of Theoretical and Applied Electronic Commerce Research |
| Subjects: | |
| Online Access: | https://www.mdpi.com/0718-1876/20/2/59 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This research aims at examining the progress of retail demand forecasting and customer classification via regression models and RFM analysis in the retail chain industry. Entailing actual retail sales data, this work utilizes three regression models:—MLP Regressor, Ridge Regressor, and KNN Regressor to forecast sales. Of them, the MLP Regressor yielded the least Mean Squared Error (MSE = 2.66 × 10) and the best coefficient of determination (R<sup>2</sup> = 0.9398) stressing its ability to identify deviations from linearity in the sales data. Also, RFM analysis, augmented by K-Means clustering, successfully categorized customers into actionable segments: loyal customers, champions, at-risk, and hibernating. Exploratory data analysis (EDA) findings indicated dramatic changes in sales and revenue, activities, and customer interactions, and products. The combined application of these approaches offers operational solutions in product acquisition, marketing communication, and revenue enhancement. The study advances current research by integrating predictive regression models with RFM segmentation, offering a dual-framework that enhances retail demand forecasting and customer behavior analysis, thereby bridging a critical gap in data-driven decision-making. However, bearing in mind that the lack of demographic data and limited feature variety may constrain the model’s ability to capture personalized customer behaviors, the findings provide a foundation for integrating more diverse datasets and advanced learning approaches for improved retail analytics. |
|---|---|
| ISSN: | 0718-1876 |