• Korean Management Science Review
• /
• v.25 no.2
• /
• pp.57-71
• /
• 2008

We study factors that affect consumers' switching behaviors among service providers in Korean mobile telecommunications service market. For empirical analysis, quarterly time series data from the first quarter of 2004 through the second quarter of 2007 were used. We chose the number of switchers to each mobile service provider in each quarter as dependent variables. Independent variables include acquisition costs per subscriber, which play the role of subsidy to mobile handset, switching costs, time trend, structural change effect, and waiting demand effects. Through the empirical analysis, we found that each provider's churn-in customers are affected by different factors. Specifically, the number of churn-in customers into SK Telecom is explained mainly by SK Telecom's customer acquisition costs and waiting demand from KTF, while the number of customers switching into KTF is better explained by switching costs from the previous service provider and waiting demand from SK Telecom. Those who chose LG Telecom as their new provider, on the other hand, were mainly attracted by LG Telecom's high subscriber acquisition cost.

• Industrial Engineering and Management Systems
• /
• v.13 no.4
• /
• pp.454-462
• /
• 2014

Retention of possible churning customer is one of the most important issues in customer relationship management, so companies try to predict churn customers using their large-scale high-dimensional data. This study focuses on dealing with large data sets by reducing the dimensionality. By using six different dimension reduction methods-Principal Component Analysis (PCA), factor analysis (FA), locally linear embedding (LLE), local tangent space alignment (LTSA), locally preserving projections (LPP), and deep auto-encoder-our experiments apply each dimension reduction method to the training data, build a classification model using the mapped data and then measure the performance using hit rate to compare the dimension reduction methods. In the result, PCA shows good performance despite its simplicity, and the deep auto-encoder gives the best overall performance. These results can be explained by the characteristics of the churn prediction data that is highly correlated and overlapped over the classes. We also proposed a simple out-of-sample extension method for the nonlinear dimension reduction methods, LLE and LTSA, utilizing the characteristic of the data.

• Journal of the Korea Safety Management & Science
• /
• v.9 no.3
• /
• pp.111-118
• /
• 2007

What is CRM(Customer Relationship Management) means that planning, executing, and re-accessing the marketing strategy based on the customer character by analyzing the material related to customers. That is CRM is a strategy of customer service on the base of data. In the case of the telecommunications and a newspaper, there are restricted application of CRM, because they are provided services by paying a given amount of money within a given period of time. This paper develops CRM model(chum prediction model) that can apply to a newspaper. For model-building, real data were used which were collected from one of the major a newspaper company in Korea. Also, this paper verifies the efficient result.

• Journal of Information Technology Services
• /
• v.10 no.2
• /
• pp.21-39
• /
• 2011

After ten years from the introduction of CRM systems, many companies reevaluate CRM from the perspective of customers. This paper classifies CRM activities at the contact points with customers and investigates the effect of CRM activities on perceived benefits and the effect of perceived benefits on behavioral intention in the Korean cosmetic industry. The result showed that contact management and complaint handling affected social benefits and that customers perceived confidence benefits from contact management, complaint handling, and churn management. Special treatment benefits were affected by all CRM activities including compensation management. Finally, all kinds of benefits had effects on behavioral intention such as recommendation intention and repurchase intention.

• Journal of Intelligence and Information Systems
• /
• v.12 no.4
• /
• pp.109-126
• /
• 2006

An effective customer classification has been essential for the successful customer relationship management. The typical customer rating is carried out by the proportionally allocating the customers into classes in terms of their life time values. However, since this method does not accurately reflect the homogeneity within a class along with the heterogeneity between classes, there would be many problems incurred due to the misclassification. This paper suggests a new method of rating customer using Holistic profit technique, and validates the new method using the customer data provided by an insurance company. Holistic profit is one of the methods used for deciding the cutoff score in screening the loan application. By rating customers using the proposed techniques, insurance companies could effectively perform customer relationship management and diverse marketing activities.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2003.10a
• /
• pp.27-30
• /
• 2003

The introduction of reseller In Australian Fixed Telecommunication Market resulted in the reduction of Telstra's local call market share by 13 percent and average fall rate was reduced. Thus, Telstra increased basic rate at 14.5 percent to compensate loss revenue in the local call market. With the deployment of carrier pre-selection of long distance and international calls, it reduced long distance rate at 23.5 percent and international tall rate at 53 percent, and increased the Churn rate. Therefore, the deployment of service-based competition brought efficient results for long distance and international call market. However, LM market created 13.4 percent reduction in call rates, complications in charge system, technical barriers and the preference of one-bill by customers.

• Information Systems Review
• /
• v.7 no.2
• /
• pp.23-40
• /
• 2005

Discovering customers' behavioral patterns from large data set and providing them with corresponding services or products are critical components in managing a current business. However, the diversity of customer needs coupled with the limited resources suggests that companies should make more efforts on understanding and managing specific groups of customers, not the whole customers. The key issue of this paper is based on the fact that the behavioral patterns extracted from the specific groups of customers shall be different from those from the whole customers. This paper proposes the idea of pre-segmentation before developing customer classification models. We collected three customers' demographic and transactional data sets from a credit card, a tele-communication, and an insurance company in Korea, and then segmented customers by major variables. Different churn prediction models were developed from each segments and the whole data set, respectively, using the decision tree induction approach, and compared in terms of the hit ratio and the simplicity of generated rules.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.39 no.4
• /
• pp.115-136
• /
• 2014

The game industry has grown steadily and the online game has become one of the most attractive game segments for its remarkable growth. Customer management in the online game industry, however, has received little attention from the academic field. The purpose of this study is to analyze the drivers of customer defection in the online game setting and suggest not only theoretical but also managerial insights into increasing customer retention rates. Prior to empirical analysis, the authors hypothesized that 3 variables of interests (Learning, Playing, Achievement) would explain the customer defection according to preceeding researches. To demonstrate these hypotheses, the authors obtained data from one of the biggest game publishers in Korea, and the empirical analysis model was developed considering context of research settings. The results of analyses provide the following insights. First, the key behavioral variables of Learning, Playing, and Achievement play substantial roles in explaining the customer defection. Next, the effects of these variables vary between customer types: novice and experienced customers. The defection decisions by novice customers are predicted by all key behavioral variables and Playing serves as the most influential indicator of the defection decisions. However, experienced customers are influenced by Playing and Achievement, while Learning has no impact on the defection decisions. Finally, the authors investigated hypothetical customer retention strategies, using the empirical results. The market outcomes indicate that the customer retention strategies work well with novice customers and it is hard-to-impossible to prevent experienced customers from defection using their behavioral data. These findings together deliver several meaningful insights to management as follow. First, the management should support customers to get involved in Learning activities at the very first stage. Second, customer's Achievement and appropriate compensation for it would work as defection barriers. Last, to optimize the outcomes of firm's marketing investments, it is better to focus on retention of novice users not experienced ones.

• Journal of Service Research and Studies
• /
• v.6 no.3
• /
• pp.57-78
• /
• 2016

The purpose of this research is to study how service recovery activities influence customers' commodity churning in the media telecommunication industry(CATV industry). Put it differently, we tried to identify this change of commodity churning rates by the stages of service failures, by which we intend to emphasize the importance of service recoveries. Korean media telecommunication market has already been saturated; customers tend to move to bigger major companies with better customer care increasingly. As once customers gone never returns, CRM functions are being reinforced over the time. We were able to have the following results. First, turning rates, for those experienced service failure, who were dissatisfied with service recovery activities are 2~5 times (monthly average turning rates are 1.3%) higher than those satisfied. Secondly, active service recovery activities at the customer's service request after experiencing service failure lowered churning rates significantly. The most effective timing is service recovery activities pre-recovery stage. Thirdly, reward activities after service recovery activities at the immediate recovery stage is more effective than service recovery at the arranged recovery schedule and reward activities after customer's expressing churning intension. The implications of this study are that firms should engage in service recovery activities at the time of identifying service failures, prior to customer's expressing churning intention, which means relatively lower ROI for the service recovery activities than the times of customers' expressing churning intention.

• Journal of Intelligence and Information Systems
• /
• v.19 no.2
• /
• pp.125-140
• /
• 2013

We call a data set in which the number of records belonging to a certain class far outnumbers the number of records belonging to the other class, 'imbalanced data set'. Most of the classification techniques perform poorly on imbalanced data sets. When we evaluate the performance of a certain classification technique, we need to measure not only 'accuracy' but also 'sensitivity' and 'specificity'. In a customer churn prediction problem, 'retention' records account for the majority class, and 'churn' records account for the minority class. Sensitivity measures the proportion of actual retentions which are correctly identified as such. Specificity measures the proportion of churns which are correctly identified as such. The poor performance of the classification techniques on imbalanced data sets is due to the low value of specificity. Many previous researches on imbalanced data sets employed 'oversampling' technique where members of the minority class are sampled more than those of the majority class in order to make a relatively balanced data set. When a classification model is constructed using this oversampled balanced data set, specificity can be improved but sensitivity will be decreased. In this research, we developed a hybrid model of support vector machine (SVM), artificial neural network (ANN) and decision tree, that improves specificity while maintaining sensitivity. We named this hybrid model 'hybrid SVM model.' The process of construction and prediction of our hybrid SVM model is as follows. By oversampling from the original imbalanced data set, a balanced data set is prepared. SVM_I model and ANN_I model are constructed using the imbalanced data set, and SVM_B model is constructed using the balanced data set. SVM_I model is superior in sensitivity and SVM_B model is superior in specificity. For a record on which both SVM_I model and SVM_B model make the same prediction, that prediction becomes the final solution. If they make different prediction, the final solution is determined by the discrimination rules obtained by ANN and decision tree. For a record on which SVM_I model and SVM_B model make different predictions, a decision tree model is constructed using ANN_I output value as input and actual retention or churn as target. We obtained the following two discrimination rules: 'IF ANN_I output value <0.285, THEN Final Solution = Retention' and 'IF ANN_I output value ${\geq}0.285$, THEN Final Solution = Churn.' The threshold 0.285 is the value optimized for the data used in this research. The result we present in this research is the structure or framework of our hybrid SVM model, not a specific threshold value such as 0.285. Therefore, the threshold value in the above discrimination rules can be changed to any value depending on the data. In order to evaluate the performance of our hybrid SVM model, we used the 'churn data set' in UCI Machine Learning Repository, that consists of 85% retention customers and 15% churn customers. Accuracy of the hybrid SVM model is 91.08% that is better than that of SVM_I model or SVM_B model. The points worth noticing here are its sensitivity, 95.02%, and specificity, 69.24%. The sensitivity of SVM_I model is 94.65%, and the specificity of SVM_B model is 67.00%. Therefore the hybrid SVM model developed in this research improves the specificity of SVM_B model while maintaining the sensitivity of SVM_I model.