Optimization of Customer Segmentation based on Campaign Post-Purchase Parameters using Evolutionary Computing Methods

Published: 2 November 2020| Version 1 | DOI: 10.17632/g3p7723r2m.1
Contributors:
Mohammad Hossein Rezvani,
Sara Hemmati

Description

Customer knowledge is one of the main issues in customer relationship management (CRM). It is vital to recognize the differences between customers, classify them, and optimally allocate resources to them according to their respective values for the company. Customer segmentation offers considerable benefits to sellers. Using this technique, one can identify and select the right customers for each marketing campaign. This is done according to the lifetime value (LTV) that the customer has generated for the company. The most important parameters that the customer provides through LTV are related to recency, frequency, and monetary (RFM) value. The purpose of the seller is to maximize the profit earned by selling a product to customers. To this end, by considering the LTV of each customer, one should calculate the optimal number of levels for each of these three parameters in such a way as to maximize the seller's profit. Given that the levels of these parameters can take any number, the number of states developed in RFM analysis is very high. Therefore, determining the optimal number for each of these parameters is an NP-hard problem. Heuristic or meta-heuristic methods are usually used to overcome this problem. In this research, we used the artificial bee colony (ABC) algorithm and the genetic algorithm based on RFM analysis to optimize the selection of the right number of customer segments for marketing campaigns. The coding was done using the R programming language.

Files

Steps to reproduce

Please see the file named "README".