Big Mart Sales Prediction using Machine Learning

INTRODUCTION: Sales prediction, also known as revenue forecasting or sales forecasting, refers to the process of accurately and timely estimating future revenue for manufacturers, distributors, and retailers, providing them with valuable insights. Sales prediction plays a crucial role in various industries, particularly in sectors such as retail, automotive leasing, real estate transactions, and other conventional businesses. OBJECTIVES: This paper focuses on developing a sales prediction model for Big Mart, a supermarket chain, using machine learning algorithms. The developed model aims to provide Big Mart with accurate sales forecasts, enabling better decision-making, improved profitability, and enhanced customer service. METHODS: The study utilises the CRISP-DM methodology and explores various machine learning algorithms, including Linear Regression, Decision Tree, Random Forest, XGBoost, Stacked Ensemble Model, and K-Nearest Neighbours (KNN). The dataset used for model development is sourced from Kaggle and includes information about products, stores, and sales. Pre-processing techniques are applied to handle missing data and feature engineering. RESULTS: The XGBoost Regression Model Tuned with RandomizedSearchCV outperforms the existing models with an RMSE of 1018.82 and an R² of 0.6181. CONCLUSION: This research contributes to the field of sales forecasting in the retail industry and provides insights for businesses looking to enhance their revenue prediction capabilities.