A Novel Approach for Enhanced Feature Selection Over Retails Sales Data Using Ensemble Machine Learning Technique

The connection between manufacturers and retailers in the retail industry, as well as supply chain management, depend heavily on sales forecasting. The efficiency of traditional methodologies and methods for completing a task has been undermined by the exponential growth of digital data. This study suggests an enhanced feature selection for retail sales using the Citadel POS (Point of Sales) Retail dataset using ensemble machine learning techniques. In order to predict sales data and provide in-depth analysis on retail sales and assessment, a variety of machine learning techniques are used for ensemble sales data. These techniques include diversified regression like Random Forest Regression, Gradient Boosting Regression, Linear Regression, and time series LSTM Model. The information used in this study was given from 2013 to 2019 by Citadel POS, a cloud-based solution that assists retail establishments in managing transactions, inventory, customers, vendors, monitor reports, manage sales, and tender data locally. The proposed method outperformed the regression and time series LSTM models with an MAE of 5.53 and an RMSE of 0.652.