taxonomy of customers identification in the banking industry using machine learning: a systematic review with a Meta-synthesis Approach

Nowadays, customers in any industry are not just looking for a product and expect to receive a personalized service based on their needs and create a different experience from the organization. From another point of view, the design of services according to the customer's needs will require a careful examination of the data related to the customer in different dimensions. Therefore, knowing the customer requires a systematic approach in order to identify the goals, influencing factors, algorithms and methods suitable for this field.

The upcoming research has analyzed the dimensions of customer recognition in the banking industry and its considerations with a data-oriented approach and the application of machine learning. The research method is applied according to the purpose and is meta-Synthesis according to the collection of information. To select the articles, 43 documents published in the period of 2016-2022 were identified as relevant and valid documents by searching in the reliable databases of Web of Science and Scopus, and further, with a meta- Synthesis approach, they were studied and coded.

The results of meta-synthesis led to the identification of three main categories: 1) Customer identification objectives understanding customer insight, identifying customer risk, organizational goals, determining customer lifetime value and product management 2) Customer identification factors: demographic, financial and behavioral and 3) machine learning algorithms: Probabilistic, Neural Networks, Ensemble, Regularization, Regression, Bayesian, Decision Tree, Dimensionality Reduction, Instanced Based and Clustering.

Based on the findings of the current research, according to the purpose of customer identification, available data and selected factors, basic and combined algorithms can be a way forward, but the important point is accurate data pre-processing. Another point is that no distinction has been made between real and legal customers, and most studies have focused on real customers, which can be attributed to the complexity of the chain of financial interactions of legal customers. Also, considering the banks' lack of emphasis on branches regarding the need to complete the information contained in the account opening forms or the lack of designing a suitable service to complete the information on electronic platforms, a more accurate understanding of the customer requires a review of these processes. It is worth mentioning that in none of the studies, customer recognition was done using only demographic factors, and depending on the purpose of the study, the factors were used in combination.