Customer Behavioral Analytics Using RFM

Lee, Kang Wenn (2019) Customer Behavioral Analytics Using RFM. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Lee Kang Wenn_Full Text.pdf Restricted to Registered users only Download (1MB)

Abstract

The main objective of this project is to extract semantic customers behavioral information through data collected from a food delivery company, Running Man. The project aims to explore customer behavior by analyzing transaction data. One of the methods applied in this project is RFM. RFM (Recency, Frequency, Monetary) is a method to profile customer into groups. RFM score calculated will be then clustered to gain more insights into the pattern. In this project, K-means clustering will be used to analyses customer purchasing behavior, and to cluster customers’ RFM score. Visuals such as RFM heat map, RFM score binning, RFM scatterplot can be used to visualize customers purchasing behavior. RFM scoring is a simple and efficient model that can be used to analyse customers purchasing behavior. Clustering RFM score can also show majority of the customers belong to one big group. The final model is able to cluster customer’s RFM score into 4 groups of customers, with each group having their unique characteristics and show insights related to the business. However, the accuracy of unsupervised learning algorithm cannot be validated. In future, more unsupervised methods can be evaluated to cluster RFM scoring. There are more unsupervised methods than k-means clustering such as mixture models, hierarchical clustering can be tested out. Binning of customer into 5 bins for every score might not produce the best accuracy thus optimization is required.