A recommender system to avoid customer churn: A case study

Abstract

A major concern for modern enterprises is to promote customer value, loyalty and contribution through services such as can help establish a long-term, honest relationship with customers. For purposes of better customer relationship management, data mining technology is commonly used to analyze large quantities of data about customer bargains, purchase preferences, customer churn, etc. This paper aims to propose a recommender system for wireless network companies to understand and avoid customer churn. To ensure the accuracy of the analysis, we use the decision tree algorithm to analyze data of over 60,000 transactions and of more than 4000 members, over a period of three months. The data of the first nine weeks is used as the training data, and that of the last month as the testing data. The results of the experiment are found to be very useful for making strategy recommendations to avoid customer churn.

Introduction

With wireless formats becoming more and more mature, wireless equipment, such as Access Point (AP), wireless network cards, etc., are getting less expensive, and mobile computing devices, such as notebooks and PDAs, becoming more popular. According to the Dell’Oro Marketing Survey Report, the market of the wireless local area network (WLAN) grew steadily about 40% year by year, from 2003 to 2006. In 2006, the net value of the global wireless networking already broke 100B dollars. The WLAN, as an extension and supplement to wired networking, will continue to be in great demand for the next generation.

When a wireless network company changes the strategy from product-based to customer-based, the customer relationship management (CRM) becomes very important. As the 80/20 rule goes, eighty percent of benefit comes from twenty percent of customers, so it is essential that companies understand customer need and develop suitable CRM strategies in order to ensure customer retention, loyalty, and satisfaction. For this purpose, companies have to collect information about customers and analyze their behavior patterns, and the ability to integrate and utilize such information effectively, therefore, is crucial to their CRM performance.

However, we have to realize customer’s behavior before we can do an excellent CRM job. To understand customer’s behavior, we need to collect all kinds of information about customers and analyze the behavior patterns of the customers to realize customer’s behavior. That would help us to adopt appropriate strategy to prevent customer loss. In fact, a regular trading system accumulated enormous amount of valuable data. Therefore, how to integrate and utilize those data would be a critical task. Data mart was the solution created especially for this issue (Demarest, 1994). Data mart is the subject to store all kinds of internal and external data for future analysis.

A recommender system is one that recommends useful information or suggests strategies that users might apply to achieve their goals. The system gives suggestions based on a given event, such as an error, or on observations of the user’s overall behavior. A simple example is a research engine that, when no results found for a query, suggests alternate keywords or queries that may achieve better results (Diamond Bullet, 2004). Recommender systems are widely used in the fields of E-commence, movies, music, books, and Web pages successfully.

Mooney and Roy (2000) suggest a content-based book recommending system that utilizes information extraction and a machine-learning algorithm for text categorization. Miller, Albert, Lam, Konstan, and Riedl (2003) propose the recommender system Movie Lens, which builds on and extends a movie recommendation (http://movielens.umn.edu) that provides movie, DVD, and VHS video recommendations, along with a search capability. Chen and Chen (2001) design the Music Recommender System (MRS), which enables a website to provide personalized music recommendation service based on music data grouping and user interest. The Group Lens recommender system helps users browse among articles in Usenet news (Konstan et al., 1997, Resnick et al., 1994). Ringo allows users to get music recommendations online and connect with other music fans (Shardanand & Maes, 1995). Again, more successful applications can be mentioned at online food stores (Svensson, Laaksolahti, Höök, & Waern, 2000), online book stores such as Amazon.com (Linden, Smith, & York, 2003), etc.