APPLICATION OF CLASSIFICATION ALGORITHM FOR SALES PREDICTION

Abstract

Increasing sales results is a desired target for all companies both at home and abroad. The company has a wide variety of products to offer. This paper (to fulfill a Business Intelligence course assignment) is the result of an experiment from data (keaggle) about consumer demand for products during the 2013-2015 period, then based on this data we try to predict to classify product sales, in order to make it easier for companies to classification for sales predictions. To find out the sales of the best-selling products, data mining classification techniques are used, namely XGBoost, Decision Tree, Random Forest, Linear Regression, and Nave Bayes. Based on the test results of the five classification techniques, the XGBoost model is the best with the data training value producing an RMSE value of 0.68% and data testing of 0.79%. This method is also better than the results of previous studies.