Abstract
Accurate market sales forecasting is crucial for effective inventory planning, pricing strategies, and resource allocation in retail operations. This study conducts a comparative analysis of various regression models—Linear Regression, Polynomial Regression, Decision Tree Regression, and Random Forest Regression—applied to the Big Mart sales dataset. The dataset, comprising 8,523 records across 12 features, undergoes a series of preprocessing steps including missing value imputation, feature encoding, scaling, and feature engineering to enhance model performance. Each model is evaluated using key metrics: Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and R² Score. The findings reveal that Linear Regression, while simple and interpretable, fails to effectively model the non-linear relationships present in sales data. Polynomial Regression shows improved accuracy by introducing non-linear feature transformations but poses a risk of overfitting. Tree-based models, especially R