مدل‌سازی منابع مطالبات معوق در بانک شهر با استفاده از داده‌کاوی حوزه استان‌های مرزی شمال غرب کشور

زمینه و هدف: با توجه به لزوم درک و شناخت بیشتر نسبت به پدیده افزایش مطالبات معوق در سیستم بانکی کشور، هدف این مطالعه ارزیابی اهمیت و شدت اثرگذاری عوامل درونی (ویژگی‌های فردی متقاضیان تسهیلات به همراه ویژگی‌های شعب پرداخت‌کننده تسهیلات) در بروز یا عدم بروز این پدیده بوده است.روش‌شناسی: برای رسیدن به این هدف با استفاده از داده‌های گردآوری تعداد 110 نفر از استفاده‌کنندگان از تسهیلات بانک شهر در 11 شعبه این بانک در استان‌های شمال غرب کشور که از طریق نمونه‌گیری خوشه‌ای-تصادفی انتخاب‌شده‌اند، در چارچوب تکنیک درخت تصمیم که یکی از مهم‌ترین تکنیک‌های داده‌کاوی می‌باشد، برای مدل‌سازی مطالبات سررسید گذشته و معوق بانکی در بانک مذکور اقدام شده است.یافته‌ها: نتایج مطالعه حاکی از آن بوده است که مدل درخت تصمیم ارائه‌شده در مطالعه، قدرت بالایی در پیش‌بینی صحیح رفتار بازپرداختی متقاضیان تسهیلات در بانک شهر به همراه داشته و توانسته‌ایم که سناریوهای مختلفی جهت بروز و یا اجتناب از بروز مطالبات معوق بانکی در بانک شهر را به همراه احتمالات آن‌ها موردبحث و بررسی قرار دهیم.نتایج: مسئولان شعبات به هنگام پرداخت تسهیلات به متقاضیانی به مولفه‌های فردی سطح تحصیلات و جنسیت مشتریان توجه داشته باشند و به ایجاد بخش اعتبارسنجی و پیگیری مطالبات معوق مستقل در هرکدام از شعبات بانک اقدام نمایند.

modeling delinquent claims in bank shahr using data mining the border provinces in the northwest of the country

background and aim: considering the need to understand more about the phenomenon of increasing overdue claims in the banking system of the country, the purpose of this study is to evaluate the importance and intensity of the influence of internal factors (individual characteristics of the applicants for facilities along with the characteristics of the branches that pay the facilities) in the occurrence or non-occurrence this has been a phenomenon.methodology: to achieve this goal, by using data collected from 110 users of bank shahr facilities in 11 branches of this bank in the northwestern provinces of the country, which were selected through cluster-random sampling, in the framework of the decision tree technique, which is one of the most important techniques. it is data mining, it has been done to model past due and overdue bank claims in the mentioned bank.results: the study's results indicated that the decision tree model presented in the study had a high power in correctly predicting the repayment behaviour of loan applicants in shahr bank, and we were able to create different scenarios for the occurrence or avoidance of overdue bank claims at shahr bank. let's discuss together with their possibilities together.conclusion: when paying facilities to applicants, branch officials should pay attention to the individual components of the level of education and gender of the customers and create an independent validation and follow-up department for pending claims in each bank branch.extended abstractintroductionnumerous field studies indicate a high ratio of non-current claims of iranian banks to their total expenditures, which is approximately three to four times higher than international standards. based on available statistics, banking delinquencies in iran have led to approximately 20% of the country's liquidity being stagnant and blocked. according to data from one of the country's banks, bank shahr, the amount of delinquent claims accounts for about 6.8%, which will have significant negative consequences. experts believe that the increasing amount of delinquent claims and past due accounts will lead to adverse effects such as weakening the value of the domestic currency, reduced public trust in the banking system and active banks, stagnation of the country's monetary resources, reduced credit capacity of the banking system, and the loss of shareholders' rights. it is natural that by reducing delinquent claims and past due accounts, the banking system can utilize its freed financial resources in other high-yield and new investment areas. this can lead to increased lending capacity, optimal management of scarce resources, better resource consumption planning, accepting commitments, and ultimately generating higher revenues (najafnejad & colleagues, 1393: 7). however, it should be noted that banking delinquencies are a complex issue influenced by various factors, including the banking network, the national network, the government, customers, their responsibility, and other contributing elements. methodologythis study is considered an applied research in terms of its objective. it also falls under the descriptive type of research, which is equipped with specific capabilities appropriate for this subject. the present research is part of modeling studies and is also a cross-sectional study regarding its time frame. the statistical population of this study consists of all individual borrowers of bank shahr in the provinces of northwestern iran, including east azerbaijan, west azerbaijan, ardebil, and zanjan (including 11 branches), whose loan installments are either current, past due, or delinquent. the sample under investigation in this study consists of 110 individuals, randomly selected from the statistical population in a cluster manner (10 individuals from each branch). to collect the required information, customer profile data from the accounting unit was obtained, including customers' identification information, loan acquisition date, customers' age, occupation, number of past-due installments, customers' gender, education level, amount of received loans, and the loan interest rate set by the bank's management for branch management in different provinces. additionally, three other independent variables, including the type of collateral received from the borrower, the age of the branch manager, and the existence or absence of a credit assessment and delinquent claims follow-up department, were collected as supplementary variables to the existing ones, based on a case study and specific branch characteristics. results and discussionthe variables of the type of collateral received from the loan applicant, the existence of an independent credit assessment and delinquent claims follow-up department in the branch, and the educational level of the loan applicant are among the most important predictive variables for determining the ability or inability of the applicant to consistently repay the received facilities. furthermore, it was observed in the study that the decision tree model showed a high capability in accurately predicting the repayment status of installments in bank shahr. therefore, it is appropriate for the bank's officials to carefully consider and implement desirable policies to prevent loan delinquencies based on the proposed scenarios. conclusionthis research, which aimed to model delinquent claims in bank shahr using data mining in the border provinces of the country's northwest, demonstrated that the designed model showed good capability in identifying and classifying the components of banking claims. hence, the method used in this research can serve as a suitable model and approach for other banks as well. to achieve this goal, after reviewing the literature and previous studies and identifying the content, dimensions, and key components governing the country's banking system, the desired model was designed and fitted with the identified variables, and internal and external important components in this bank were identified and compared with similar studies to ensure validity.thus, in general, the following policy recommendations can be presented to the officials of bank shahr to prevent loan delinquencies:branch officials should pay more attention to the type of collateral received when providing loans to applicants with bachelor's degrees or lower education and requesting loan amounts over 200 million rials.branch officials should also pay more attention to the type of collateral received when providing loans to female applicants with high loan amounts.bank shahr officials should give more consideration to establishing independent credit assessment and delinquent claims follow-up departments in each branch of the bank. fundingthere is no funding support. authors’ contributionall of the authors approved thecontent of the manuscript and agreed on all aspects of the work. conflict of interestauthors declared no conflict of interest. acknowledgmentswe are grateful to all the scientific consultants of this paper.