On the eve of Black Friday, goods are sold at a very high discount and thus increases the sales around 30 folds as compared to the normal flash sale days. The customer's data those have bought products on this day can be analyzed which will give a brief declaration of their choice on various products. Here, we have analyzed the data containing the tuples of customers along with their buying factors and amount. This data is analyzed and predicted just to provide special discounts on goods for the customers according to their taste and buying budget. Four models that are xgboost, tfidftransform, both combination and extra trees regressor have been used for the prediction for different distributions of training and testing data (50:50, 70:30, 30:70) and for a separate sample training dataset and testing dataset which includes further two cases of prediction. These two cases consist of predicting and testing on the same dataset and predicting on the training dataset and testing it on a separate testing data set. Analysis is done (Exploratory data analysis) on the dataset to understand the customer behavior and patterns of various products popularity during the sale. Feature importance and gain importance are shown for all five cases. The accuracy results are highlighted in the form of root mean square error (RMSE) score and the accuracy of all the models in different cases have been depicted in the form of their accuracy graphs. © 2020 IEEE.