This research paper provides an algorithm which deals in sales forecasting to enable sales organizations to make business decisions and predict sales forecast for short-term and long-term performance. This research work is to plan a model to foresee the stock through Machine Learning. In this, we have old sales data of products of the stores and from that sales data, we will predict the future sale of items. This research is carried out as companies can base their forecasts on the previous sales data. The target of this research work is to improve forecasting practices. The hybrid approach sales prediction is presented to get more accurate results. This work has been implemented in Python using Machine Learning models. Different models are implemented using data set available on the Kaggle.com, for research work to be done by the researchers for better- predicted resultsthe results are accurate due to the hybrid approach of Auto Regression Moving Average, Rolling Forecast and Exponential Smoothing algorithm as tested on the data available for 2017. The goal of this research work aims the performance of the organization—more profit, more revenue, financial improvements. The result introduced in this research recommends that by utilizing the hybrid approach, forecasting will deliver
request figures that are considerably more precise than those given by the well-known algorithm.