Big Mart Sales Prediction Using Machine Learning

This paper developed a prediction model that will forecast product sales at a particular shop using numerous datasets. This study is able to get findings with a required degree of accuracy using the method employed to create a comprehensive model. Additionally, this information can be used to take decisions to additionally foster arrangements. This paper proposed many issues to predict sales of Big Mart and how to predict by employing five different types of regression algorithms to forecast it, including XGBoost regressor, linear regression, decision trees, random forest regressor, and grid search CV. These algorithms were employed to build models, compared the correctness of those models, and prepared them to fit the board's assumptions so that cautious actions could be done to accomplish the affiliation's purpose. These models may be applied in many districts. It has also been addressed how to anticipate how different kinds of items will be arranged and how different conditions will affect those arrangements. In this research, there are many implementation challenges in practical models. Various models are created and after comparing their R2-Score the best R2-score is of random forest grid search which is equal to 0.5588858290914282. Somehow R2-score is low, but it is due to the variability and lack of data, and it is good enough to predict the output.