A Comparative Study of Sales Prediction Using Machine Learning Models: Integration of PySpark and Power BI

Retail management requires accurate sales forecasting for strategic planning, inventory control, and revenue maximisation. In this research, we are predicting sales using PySpark-built ML models and the Big Mart dataset. We evaluate and demonstrate the prediction skills of numerous machine learning algorithms, concentrating on XGBoost. Big Mart offers item weight, visibility, type, and outlet data. We use these properties as prediction model features. PySpark, a strong distributed computing platform, manages massive datasets for analysis and model training. In massive trials, we test decision trees (DT), XGBoost, linear regressions (LR), and random forests (RF). Training and testing enhance model accuracy. RMSE and R-squared measure our model quality. Our metrics evaluate the model’s data fit and prediction accuracy. XGBoost performed best with an RMSE of 1081 and an R-squared of 0.59. The XGBoost algorithm accurately predicts Big Mart sales. The model performs well because of its ensemble learning and understanding of intricate dataset links. We also used Power BI to present analytical insights, helping decision-makers design sales-estimated business plans. This study employed several machine learning algorithms, XGBoost gave the best performance. This research provides insights into how organisations might use these technologies to improve resource allocation and inventory management decisionmaking.

Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.