Amir Khorrami; Mohammad Taghi Taghavifard; Seyed Mohammad Ali Khatami Firouzabadi
Abstract
Credit risk assessment is one of the key issues for banks and financial institutions and various models have been developed for this. This study uses Case Based reasoning (CBR) Model and considers a database of bank credit customers to assess the credit risk of bank applicants. For this, 9 criteria were ...
Read More
Credit risk assessment is one of the key issues for banks and financial institutions and various models have been developed for this. This study uses Case Based reasoning (CBR) Model and considers a database of bank credit customers to assess the credit risk of bank applicants. For this, 9 criteria were selected based on the experts' opinion and were weighted using the Fuzzy Analytical Hierarchy Process (FAHP). Return check, housing situation and income level are the most important criteria for credit risk assessment of the bank applicants. Then, using the TOPSIS Technique, we could evaluates the similarity of the new item with actual past cases or evaluate the new applicant with the ideal option, and uses a case-based reasoning model to predict the likelihood of default or non-default applicants. Survey research was applied for this study and the research community was the records of previous bank applicants between 1390-94 years. This research is an applied and descriptive and descriptive study. The results show that the accuracy of the CBR model is higher than other validation and ranking methods of bank customers. The use of the CBR model in order to authenticate customers has obtained results far better than the performance of the credit sector experts, which led to the judgment of default or non-default of customers, indicating the high performance of the model used in comparison to the model used by bank and validation experts. CBR leads to the design an expert, specialized and intelligent system which addition to storing data in a database, stores models and templates for use.
Tahereh Zaefarian; Mohammad Andabili; Hossein Momeni; Seyed Esmaeil Najafi
Abstract
Today, there are more than 300 types of cars in Iran auto market, which has a significant growth in recent decade. High variety have challenges for decision makers in selecting cars. No mathematical model has been developed yet for segmenting and ranking Iran auto market, which carry out both defining ...
Read More
Today, there are more than 300 types of cars in Iran auto market, which has a significant growth in recent decade. High variety have challenges for decision makers in selecting cars. No mathematical model has been developed yet for segmenting and ranking Iran auto market, which carry out both defining automatic cluster numbers as well as automatic weighting criteria by the model.This research develops a Hybrid DEMATEL-Two-Step Clustering-TOPSIS approach. The model first finds the beat appropriated criterion for segmentation. Then uses a two-step clustering approach for segmenting Iran auto market based on price criterion. Second, the criteria will be weighted automatically using Shannon entropy weighting method and then, TOPSIS method rank competitors in each defined price segment (lower 900 Million Rials). Also, the Spearman's rank correlation test is used to compare the model results with Iranian customer behavior (with selling volume). The price segmentation results reveal that the Iran auto market can be segmented in six different levels. Furthermore, the ranking results disclose that price is not the only effective factor in finding car utility for the buyer. A weighted combination of performance, features and price will determine optimized selection for buyers
Siamak Kheybari; Mostafa Kazemi
Abstract
TOPSIS is located in compensatory decision-making methods. The basic principle is that the chosen alternative should have the shortest distance from the positive ideal solution and the longest distance from the negative ideal solution. The existence of incremental and decreasing uniform trend in positive ...
Read More
TOPSIS is located in compensatory decision-making methods. The basic principle is that the chosen alternative should have the shortest distance from the positive ideal solution and the longest distance from the negative ideal solution. The existence of incremental and decreasing uniform trend in positive and negative criteria is one of the TOPSIS characteristics in determining positive and negative ideal points. While the utility assigned to the decision maker after a certain level in each criterion will be less tangible. Therefore, in the ranking of alternatives in addition to the value of each alternative in each indicator, the balance among criteria of each alternative should be considered so the alternative that has an appropriate place in an important indicator, but does not have an appropriate place in other criteria, not to be the first chance for selecting. For this purpose, in this paper by adding virtual dimension to the decision matrix that comes from the deviations among criteria of each alternative, we have tried to compensate the mentioned weakness. To evaluate the proposed method, three different examples are presented. Thus, each of the three provided examples, solved by proposed method, TOPSIS, VIKOR, Deng and SAW and then by using the Spearman correlation coefficient the number of significant correlation relationships between the proposed method and TOPSIS with the other three methods were compared. Then, the rating similarity percentage of the proposed method and TOPSIS were compared with VIKOR, Deng and SAW.
Emad Roghanian; Fatemeh Mojibian
Volume 10, Issue 26 , January 2012, , Pages 35-54
Abstract
In this paper, a novel intuitionist fuzzy TOPSIS method for groupdecision making will be presented. In this method the preferencevalues for an alternative on criteria and the weight values of criteriaare given by experts, using linguistic values of trapezoidal intuitionistfuzzy numbers, and weights of ...
Read More
In this paper, a novel intuitionist fuzzy TOPSIS method for groupdecision making will be presented. In this method the preferencevalues for an alternative on criteria and the weight values of criteriaare given by experts, using linguistic values of trapezoidal intuitionistfuzzy numbers, and weights of decision makers’ opinions areunknown. In proposed method, expected values and weightedaveraging operator for trapezoidal intuitionist fuzzy numbers are usedto induce the weight values of criteria and decision makers’ opinions.Then an algorithm for ranking alternatives is presented undertrapezoidal intuitionist fuzzy environment. Finally, using a numericalexample, the efficiency of new extended TOPSIS method isinvestigated.