Enhancing Sales Forecasting Accuracy through DBSCAN Clustering and Ensemble Modeling Techniques

Abstract

This study aims to enhance sales forecasting accuracy by integrating clustering techniques with ensemble predictive modeling. The primary objectives include identifying distinct sales patterns and developing a robust forecasting model that leverages these insights. The analysis utilized a dataset of weekly sales transactions, employing the DBSCAN algorithm for clustering to uncover underlying sales patterns. Subsequently, various regression techniques, including Linear Regression, Random Forest Regression, and Gradient Boosting Regression, were applied. The results from these models were integrated into an updated ensemble model, which demonstrated improved predictive performance. The ensemble model achieved a Mean Absolute Error (MAE) of 0.516 and an R-squared value of 0.993, significantly outperforming traditional regression models. The clustering results, visualized through Principal Component Analysis (PCA), provided valuable insights into customer behavior and sales trends, allowing for more accurate forecasts. These findings suggest that integrating advanced analytics into sales forecasting can lead to better strategic decision-making. This study underscores the significance of combining clustering and ensemble modeling techniques in sales forecasting. By capturing complex sales patterns and improving predictive accuracy, organizations can optimize their operational strategies and enhance overall business performance. The research contributes to the growing body of literature on machine learning applications in sales forecasting, highlighting the importance of innovative approaches in a competitive market environment.