Optimized Predictive Model Using Artificial Neural Network for Market Basket Analysis

Abstract

Market Basket Analysis (MBA) is a modeling technique in view of the theory that in the event that you purchase a specific items, you are progressively likely to purchase another items. The changing requests of the consumer with estimation of seasons are the main task against the market basket analysis. MBA is nothing but predictive model which is used to predict the buyer’s behaviour with goal of finding the relationship among various products from their market basket. The optimization in finding of such relationships can help the retailers and merchants to design a sales strategy by considering the items frequently purchased together by customers. Regardless of benefits of using MBA, there are some major research challenges associated with the MBA designing in previous methods. As there is significant growth of online shopping portals and product purchase now days, the current predictive models are ineffective and inefficient over large sales datasets. In this paper, we are attempting to design optimized predictive model to overcome the current research problems. We proposed novel predictive model for MBA by using data cleaning and neural network approach. Our designed data cleaning method helps to improve the quality of input dataset and hence MBA results by removing the all types of errors from it. Secondly unsupervised machine learning based MBA model based on artificial neural network designed. The existing Apriori algorithm is modified by using neural network method in order to optimize the prediction results. To the best of our knowledge, this is the first attempt in MBA. The practical results showing that proposed predictive model for MBA outperforming the previous method