LSTM Based Sales Prediction System for Supermarket

Many supermarkets today do not have a better forecasting system for yearly sales. This is due to the lack of skills, resources, and knowledge to make sales estimations. The use of the conventional statistical technique to forecast grocery supermarket sales has left many demanding situations unaddressed and, in general, results in the creation of predictive models that perform poorly. The era of big data, coupled with access to massive computing power, has made deep learning a go-to for sales forecast. This proposed system investigates forecasting sales for several stores. The essential variables that helped in better sales forecast were store type, date, item and sales. Product sales forecasting is a significant aspect that mainly aims to manage the purchase. Forecasts are crucial, especially in determining the accuracy of estimating the future demand for goods and inventory stock levels, which has been vital, mainly in the Grocery or supermarkets. Consider if goods are not available readily or the availability of goods is more than the demand, it can compromise overall profit. As a result, product sales forecasting may be critical to ensure that losses are reduced. Besides, the issue is becoming complicated for retailers to add specific consumer criteria that involve new, ever-changing seasonal preferences and volatile product marketing items. Hence a forecasting model is developed using Long short term memory (LSTM) to improve product sales accurate forecasts. The proposed model is mainly targeted to support the future purchase and a more precise prediction of sales and is not meant to alter Forecasting's existing subjective methods. A model based on a real grocery store's data for four years is developed to predict the store's sales. This method's impact is to forecast the availability of products in stores to ensure that they have just enough products at the right time.