Mohsen Asgari; Mohammadreza Taghva; Mohammad Taghi Taghavifard
Abstract
Banks are endeavoring to gain more funds in a highly competitive environment. Given the higher costs of attracting new customers than retaining existing ones, most banks focus on maintaining their existing customers. Therefore, it is quite important for the banks to predict the customer churn in advance. ...
Read More
Banks are endeavoring to gain more funds in a highly competitive environment. Given the higher costs of attracting new customers than retaining existing ones, most banks focus on maintaining their existing customers. Therefore, it is quite important for the banks to predict the customer churn in advance. In almost all related research works in banking, customers are divided into two types of static categories: “churners” and “loyal” customers. However, due to the nature of banking particularly in Iran, it is necessary to define churn in a dynamic manner in a variety of circumstances. In this study, the concept of state chain is introduced, which identifies changes in customers’ partial churn status over time. Using the sequence of chains and a combination of hierarchical clustering techniques as well as support vector machine, a model was developed to predict partial churn of bank customers. To construct a practical sample and to evaluate the prediction accuracy, 5 years of real European bank customers’ data as well as 3 years of customers’ data from three different Iranian banks were used. The results indicate a high level of prediction accuracy for the model in all 4 banks, particularly when longer sequences of states are used.
Mohammad Reza Taghva; Mohammad Taghi TaghaviFard; Ali Moeini; Mohammad Reza Zynoddini
Abstract
For about two decades, the different organizations have begun offering various types of electronic services in the world and many researchers have spoken about governments using multiple approaches to become smarter. However, there is no consensus and no precise definition of the smart government, ...
Read More
For about two decades, the different organizations have begun offering various types of electronic services in the world and many researchers have spoken about governments using multiple approaches to become smarter. However, there is no consensus and no precise definition of the smart government, its components and dimensions. On the other hand, the high cost of implementing macro-scale e-service projects encourages governments to initially model e-government and control costs. Hence, over a decade, various models for the development, delivery, evaluation, and other dimensions of e-government have been presented and discussed, but because of the novelty of the smart government as the new generation of e-government, no specific framework and model are provided. The purpose of this research is to identify the components and dimensions of smart government and their classification. Therefore, it tries to use the meta-synthesis method to systematically study scientific documents related to the discussion and to express the dimensions of the smart government. Finally, by using the Shannon entropy method, the indices of each dimension are ranked according to previous studies. The result of the research shows that the smart government has six dimensions: 1. Smart management and leadership; 2. Infrastructure and smart technology; 3. Smart interaction; 4. Smart service; 5. Smart environment; and 6. Smart security. The results of the research show that the proposed model is more complete than other studies, and it has a specific category. This research is the first research to rank the indicators and dimensions of smart government.