Mohammad Taghi Taghavifard; Ahmad Nadali
Volume 9, Issue 25 , July 2012, , Pages 85-107
Abstract
This research study aims at using Data Mining and Fuzzy Logicapproaches to classify the credit scoring of banking system applicantsas to cover uncertainties and ambiguity connected with applicantclasses and also variables that affect their behavior.The methodology, according to a standard Data Mining ...
Read More
This research study aims at using Data Mining and Fuzzy Logicapproaches to classify the credit scoring of banking system applicantsas to cover uncertainties and ambiguity connected with applicantclasses and also variables that affect their behavior.The methodology, according to a standard Data Mining process, is tocollect and refine the client data, then those variables which are inlinguistic forms are converted to fuzzy variables under the supervisionof banking experts and final data are modeled using Fuzzy DecisionTree, subsequently. The unfuzzy data are also modeled using the otheralgorithms.The results of the study suggest that as far as client distinctionaccuracy is concerned Fuzzy Decision Tree produces better resultscompared to Traditional Trees, Neural Networks, and statisticalprocedures such as Logistic Regression and Bayesian Network.However, it is not as accurate as Support Vector Machine and GeneticTree. On the other hand, Fuzzy Decision Tree technique has gainedbetter prediction than prediction performance of bank credit scoringexperts.
Saviz Mohammadnabi; Sina Mohammadnabi
Volume 9, Issue 24 , March 2012, , Pages 161-182
Abstract
This study attempted to use data mining as a powerful analytical tool to find patterns for occurrence of accidents from 1845 recorded events in safety data warehouse in one of the largest project-based organizations active in construction industry in Iran between the years 2002 and 2008. High-risk nature ...
Read More
This study attempted to use data mining as a powerful analytical tool to find patterns for occurrence of accidents from 1845 recorded events in safety data warehouse in one of the largest project-based organizations active in construction industry in Iran between the years 2002 and 2008. High-risk nature of construction industry, Geographic expansion of the projects sites and large number of accidents are the characteristics of this organization. Predicting and preventing models for occurrence of accidents have been proposed in this study by extracting 31 traceable Association rules from recorded events. Extracting the rules, the minimum amount of confidence, support and lift indicators have been set respectively in 73%, 5% and 1 levels.
M.J. Tarokh; K. Sharifiyan
Volume 6, Issue 17 , September 2007, , Pages 153-181
Abstract
Financial corporation and banks are sort of organization that due to specialty of their work, are very needy to customer management process ; and data mining is one of the best available tools for them to asses definition and behavior forecast of their customers.
Data mining is improving very fast and ...
Read More
Financial corporation and banks are sort of organization that due to specialty of their work, are very needy to customer management process ; and data mining is one of the best available tools for them to asses definition and behavior forecast of their customers.
Data mining is improving very fast and due to presence of vide range of data using computer is essential. Nets & powerful algorithms are used to emplace of manual analysis to derive knowledge & information from data.
In this paper: “Mellat Bank” and its information bank of different division has been evaluated after data extraction from information bank and noise distortion , k means algorithm and fuzzy - k - means algorithm standard test of cluster's compression were used for customer clustering in groups. Determination of optimum number of clusters is done by applying cluster quality assay function. Afterward was used to determine the quality of gained clusters. Then the value of each cluster was determined through FRM model. At the end of project for clusters analysis and define appropriate strategy for each cluster; the pyramid of customer value was used.
