Retail Sales Prediction Using Machine Learning Algorithms

Abstract

Data and Data Science, both are playing a pivotal role in creating business intelligence. Machine learning became buzzword technology in data processing and analytics. Retailors in an unorganized sector take decisions intuitively whereas organized retail concerns use business intelligence for their business decisions. Retail sales prediction is common in the organized retail sector, which is highly useful for in time and strategic competitive decisions. In general, sales predictions are made based on past data using statistical and mathematical techniques, etc. The present paper is an attempt to predict retail sales using machine learning algorithms and to present the accuracy of results to the retailors, managers, and policymaker in the retail industry. In this paper three machine learning algorithms k-Nearest Neighbour regression model, Multinomial regression model, Ada Boost regression model were implemented on the training dataset. Multinomial and Decision Tree with Ada Boost regression models work with 100% accuracy. Decision Tree with Ada Boost regression model was implemented on the test set to predict the outlet sales. The regression model predicted the sales of a particular item indicated with the item identifier in a particular outlet indicated with the outlet identifier on the test set. The implications for retailers, managers, and policymakers in the retailing sector are provided based on the results