Research trends in customer churn prediction: A data mining approach

• Autor(a) principal: Tianyuan, Z.
• Data de Publicação: 2021
• Outros Autores: Moro, S.
• Idioma: eng
• Título da fonte: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Texto Completo: http://hdl.handle.net/10071/23655
• topic: Telecom; Data mining; Customer churn prediction

This study aims to present a very recent literature review on customer churn prediction based on 40 relevant articles published between 2010 and June 2020. For searching the literature, the 40 most relevant articles according to Google Scholar ranking were selected and collected. Then, each of the articles were scrutinized according to six main dimensions: Reference; Areas of Research; Main Goal; Dataset; Techniques; outcomes. The research has proven that the most widely used data mining techniques are decision tree (DT), support vector machines (SVM) and Logistic Regression (LR). The process combined with the massive data accumulation in the telecom industry and the increasingly mature data mining technology motivates the development and application of customer churn model to predict the customer behavior. Therefore, the telecom company can effectively predict the churn of customers, and then avoid customer churn by taking measures such as reducing monthly fixed fees. The present literature review offers recent insights on customer churn prediction scientific literature, revealing research gaps, providing evidences on current trends and helping to understand how to develop accurate and efficient Marketing strategies. The most important finding is that artificial intelligence techniques are are obviously becoming more used in recent years for telecom customer churn prediction. Especially, artificial NN are outstandingly recognized as a competent prediction method. This is a relevant topic for journals related to other social sciences, such as Banking, and also telecom data make up an outstanding source for developing novel prediction modeling techniques. Thus, this study can lead to recommendations for future customer churn prediction improvement, in addition to providing an overview of current research trends.