Investigating Factors Impacting Customer Churn in Banks: A Comparative Study of ML Models and Multimodal Fusion
Main Article Content
Abstract
In today's competitive business landscape, organizations strive to enhance their quality of service (QoS) to meet the increasing demands of customers. Customer Relationship Management (CRM) systems play a vital role in acquiring new customers, establishing lasting relationships, and improving customer retention for sustained profitability. By leveraging machine learning models, In order to achieve a competitive advantage, customer relationship management systems are able to analyze the personal and behavioral data of customers accurately predicting customer churn and identifying the underlying reasons. This paper aims to predict customer churn, analyze the associated factors, and provide insights for improvement. These predictions enable organizations to design targeted marketing plans and service offerings.
a) a) Four separate analytic approaches from different types of learning are chosen for this study so that the performance of different machine learning techniques for churn prediction can be compared and analyzed. Ensemble-based (Random Forest) and multi-model (XgBoost) and machine-learning (SVM) methods. These techniques are applied to a dataset containing 10,000 records of Bank Customer Data.
The results of the analysis demonstrate that XgBoost achieve a high accuracy rate of 85% for customer churn prediction. Additionally, Random Forest, ANN, Support Vector Machines and Multimodal Algorithm exhibit promising accuracy rates of approximately 86% and 87% respectively. The findings highlight the effectiveness of these machine-learning techniques in predicting customer churn.
Furthermore, the analysis of the churn predictions reveals noteworthy insights. Specifically, it is observed that customers in the age group of 50-60 exhibit a higher churn rate compared to the retained customers. Additionally, if precautionary measures are not taken, this trend is expected to continue for the age group of 40-50. Moreover, the analysis indicates that customers with credit cards are more prone to churn compared to those without credit cards, emphasizing the importance of credit card-related strategies in reducing churn.
a) a) Four separate analytic approaches from different types of learning are chosen for this study so that the performance of different machine learning techniques for churn prediction can be compared and analyzed. Ensemble-based (Random Forest) and multi-model (XgBoost) and machine-learning (SVM) methods. These techniques are applied to a dataset containing 10,000 records of Bank Customer Data.
The results of the analysis demonstrate that XgBoost achieve a high accuracy rate of 85% for customer churn prediction. Additionally, Random Forest, ANN, Support Vector Machines and Multimodal Algorithm exhibit promising accuracy rates of approximately 86% and 87% respectively. The findings highlight the effectiveness of these machine-learning techniques in predicting customer churn.
Furthermore, the analysis of the churn predictions reveals noteworthy insights. Specifically, it is observed that customers in the age group of 50-60 exhibit a higher churn rate compared to the retained customers. Additionally, if precautionary measures are not taken, this trend is expected to continue for the age group of 40-50. Moreover, the analysis indicates that customers with credit cards are more prone to churn compared to those without credit cards, emphasizing the importance of credit card-related strategies in reducing churn.
In conclusion, this research paper underscores the significance of accurate churn prediction for organizations in their quest to enhance QoS and improve customer retention. The comparative analysis of various machine learning techniques provides valuable insights into their performance, with Random Forest and Support Vector Machines showcasing superior accuracy rates. The identified trends related to age groups and credit card usage further inform targeted marketing efforts and service enhancements to mitigate customer churn.
Article Details
Section
Articles