Design and Implementation of Sales Prediction Model Using Decision Tree Regressor over Linear Regression Towards Increase in Accuracy of Prediction

Abstract

The purpose is to predict future price of product and assists companies to make business strategic plans to increase overall sales and also an experiment is performed to find the best suitable algorithm among Linear Regression and Novel Decision tree regressor. Predicting future price of a product using linear regression algorithm (N=10) and Novel Decision tree regressor (N=10). Dataset used is Bigmart Sales data from kaggle. The sample size is 542 for each group. Novel Decision tree regressor produces a better accuracy of 97.5% and for Linear regression classifier is 87.6%with a statistical significance value of p is 0.03 (p>0.05). The results proved that the Novel Decision tree regressor is significantly better for sales forecasting than linear regression algorithm within the study’s limits.