Regressor Based Supermarket Sales Prediction Using Time Series Data

In a variety of problem domains, including sales, finance, healthcare, the stock market, etc., forecasting techniques are applied. The supermarket sales forecast aids in increasing sales in a professional setting. The method aids in problem-domain decision-making. Most supermarket chains estimate and evaluate sales for the following year using, at most, ad hoc techniques and technologies. Many issues remain unresolved as a result of using traditional statistical methods to anticipate grocery sales. Supermarkets are in need of accurate sales forecasts. Most supermarket chain stores, use informal strategies and methodologies to analyze and forecast sales. This is due to a lack of knowledge about optimizing the prediction result. In order to solve this problem, this study creates the most accurate model for forecasting product sales at the supermarket based on training and testing results. Three models are created in this study to forecast product sales in a supermarket. These models, which forecast product sales in supermarkets across branches, are trained using historical supermarket data collected in diverse places. It investigates and compares the three machine learning models—Random Forest, Decision Tree, and Linear Regression for forecasting supermarket sales. This makes it easier for supermarkets to forecast product sales and helps them turn a profit. To identify the most accurate model, the prediction errors are computed.