Research on the Distributions of Products for Big Mart

Abstract

In order to have a brief insight into the process of business data analysis for the big marts product and through which to find out the inner logic about data analysis. This research did a brief research based on the big mart sales dataset from Kaggle. The data are collected in 2013 for 1559 products across 10 stores in different cities. This research aims to build a predictive model and forecast the sales of each product at the specific stores and then try to understand the properties of products and outlets which play a key role in increasing sales. After using some basic analysis methods based on python, the author gets the distribution outcome of a big marts product and creates five simple models to predict the final outlet-sales and find out the most performed model using MAE criteria. The outcome shows that finally the XGB Regressor model performed best and for the real business, it is the most suitable selection.