Market Basket Analysis for Next Basket Item Prediction Using Data Mining and Machine Learning

Abstract

Data Mining is one of the morst prominent approach used nowadays to identify sales patterns and features from large scale datasets. The primary objective of this research is to develop a model based on advaned Market Basket Analysis (MBA) to increase the sales of any orgnaization. This research focused on adoption of FP-Grwoth algorithm and Machine Learning (ML) algorithms to predict next item basket. Two widely used datasets French Retail Store Dataset (FRSD) and Bread Basket Dataset (BBD) were used for experiments. Experiments showed that FP-Growth algorithms produced most frequent items purchased by customers while ML classifiers such as Logistics Regression (LR), Random Forest (RF), K-Nearest Neighbors (KNN) and Decision Tree (DT) showed promising results. Among these ML classifiers RF produced promising accuracy of 0.922% and 0.930% using FRSC and BBD respectively. The proposed model has ability to predict next item basket. This model will help organization to increase their sales.