Analyzing customer behaviours and predicting an optimal credit limit

Abstract

Analyzing customer behaviours to gain many vital insights is a prominent topic in different industries. The main reason for such analysis is that it helps the businesses to identify valuable insights that could uplift both company profit and customer satisfaction. This research focuses on such area where both customer satisfaction and company profit could be uplift by identifying customer behaviours. The selected customer base is an LTE broadband customer base of a telecommunication company. The objective of this research is to predict an optimal credit limit for customers by analyzing different customer behaviours benefiting both the customer and the company. In order to identify different customer behaviors in terms of their payment and data usage patterns, clustering algorithms were used. This research discusses different cluster quality indexes to measure the goodness of the clusters. Once the customers are clustered into different groups, regression models were used to derive customized formular to identify the relationship between the profit generation, credit limit and the other features. Then dynamic programming is used to identify the optimal credit limit for each group of customers identified. Further to this a classification model is used to clarify the customers to relevant clusters identified in the future. The outcome of this research shows two different customer behaviours resulted with two different credit limits.