• OpenAccess
    • List of Articles Data mining

      • Open Access Article

        1 - Applying data mining techniques to regions segmentation for entrance exams to governmental universities
        نرجس سرعتی آَشتیانی somayyeh alizadeh علی  مبصّـری
        The large numbers of Iranian high school graduates are willing to enter in governmental and popular colleges and compete for it. On the other hand, these graduate students are from various regions with different levels of access to facilities. In opinion of directors of More
        The large numbers of Iranian high school graduates are willing to enter in governmental and popular colleges and compete for it. On the other hand, these graduate students are from various regions with different levels of access to facilities. In opinion of directors of relevant agencies, the quota allocation solves this problem and they are looking to use the knowledge hidden in the data are available in this area.By this way volunteers from each region are compared together and managers are helped to allocate proper quota to related students in regions of each segment. In recent years, quota allocation was determined by Taxonomy that its result is a kind of ranking that does not allow group analyzing and identifies number of region theoretically. To solve this problem clustering is a good strategy. This study is carried out by using data mining techniques and Crisp methods on related dataset from education ministry, interior ministry, ministry of health, and center of statistic and evaluation organization for the first time. After extracting of effective attributes in this area, data preparation, data reduction and combination of attributes using Factor Analysis have done.in next step, by using K-means algorithm, similar items assign in to a cluster that has the minimum distance with centroid mean and then by using neural networks and decision trees, new item can be devoted to each cluster. Finally for assessing created models, accuracy of outputs compared with other methods. Outcomes of this research are: determining the optimal number of sectors, segmenting regions, analyzing each section, extracting decision rules, predicting class labels for new areas faster and more accurately, allowing the appropriate strategies formulation for each section Manuscript profile
      • Open Access Article

        2 - Survey different aspects of the problem phishing website detection and Review to existing Methods
        nafise langari
        One of the latest security threats in cyberspace to steal personal and financial information is created by phisher. Due to there Are various methods to detect phishing and also there is not an up-date comprehensive study on the issue, the authors Motivated to review and More
        One of the latest security threats in cyberspace to steal personal and financial information is created by phisher. Due to there Are various methods to detect phishing and also there is not an up-date comprehensive study on the issue, the authors Motivated to review and analysis the proposed phishing detection methods in five categories such as: anti-phishing tools Based, data mining based, heuristic based, meta-heuristic based and machine learning based methods. The advantages and Disadvantages of each method are extracted from the current review and comparison. The outlines of this study can be suitable to identify the probability gaps in phishing detection problems for feature researches. Manuscript profile
      • Open Access Article

        3 - Integrating data envelopment analysis and decision tree models In order to evaluate information technology-based units
        Amir Amini
        In order to evaluate the performance and desirability of the activities of its units each organization needs an evaluation system to assess this desirability and it is more important for financial institutions, including information technology-based companies. Data enve More
        In order to evaluate the performance and desirability of the activities of its units each organization needs an evaluation system to assess this desirability and it is more important for financial institutions, including information technology-based companies. Data envelopment analysis (DEA) is a non-parametric method to measure the effectiveness and efficiency of decision-making units (DMUs). On the other hand, data mining technique allows DMUs to explore and discover meaningful information, which had previously been hidden in large databases. . This paper presents a general framework for combining DEA and regression tree for evaluating the effectiveness and efficiency of the DMUs. Resulting hybrid model is a set of rules that can be used by policy makers to discover reasons behind efficient and inefficient DMUs. Using the proposed method for examining factors related to productivity, a sample of 18 branches of Iran insurance in Tehran was elected as a case study. After modeling based on advanced model the input oriented LVM model with weak disposability in data envelopment analysis was calculated using undesirable output, and by use of decision tree technique deals with extracting and discovering the rules for the cause of increased productivity and reduced productivity. Manuscript profile
      • Open Access Article

        4 - Discovering spam in Facebook social network using data mining.
        amin nazari
        In recent years, by developing new technologies and communication facilities such as internet, new aspects named virtual social networks have been created. Rapid development of social networks and huge number of anonymous Users in these networks, created a suitable en More
        In recent years, by developing new technologies and communication facilities such as internet, new aspects named virtual social networks have been created. Rapid development of social networks and huge number of anonymous Users in these networks, created a suitable environment for scammers. Most of the times, scammers are trying to spread several types of spams into these high potential places. Hence, an effective method is required to detect the spams in order to increase the level of information security of people in the social networks. In this paper, a new method for discovering spammer in Facebook social network is proposed. Findings show 99.96% accuracy. In previous papers, users were divided into two groups of ordinary users and spammer users. The method of classification in these papers recognizes also as a spam the users which attached by spammer. So, in this paper by dividing users into three types of ordinary users, spammer and users attached by spammer, accuracy of spam detection has been increased. Manuscript profile
      • Open Access Article

        5 - Developing A Suitable Data Model For Data Mining Application In Banking
        Shahideh Ahmadi
        Banking domains such as credit assessments, branch efficiency, electronic banking  is tremendous contexts for the broad application of the concepts of business intelligence and its methods include data mining, data warehouses and decision support systems. There are many More
        Banking domains such as credit assessments, branch efficiency, electronic banking  is tremendous contexts for the broad application of the concepts of business intelligence and its methods include data mining, data warehouses and decision support systems. There are many researches in the field of application of data mining in particular domains of banking, each of which analyzes the different entity of the banking sector, such as customers, facilities, accounts, and so on, but there isn’t research that comprehensively addresses all data mining applications in a bank, it integrates them, extracts and categorizes all banking entities for a variety of analytical applications and ultimately provides an appropriate data model according to the required attributes for the banking domains. Currently, information systems of Iranian banks are being developed for responding to new information needs. In this research by using content analysis method was investigated the content of valid research in the field of banking which was carried out with the data mining approach and by extracting the entities and attributes used in these researches is presented an appropriate data model for data analysis applications in banking. Information technology managers  by using this model can assess the status of the bank in terms of the richness of the data needed to conduct data analysis and consider the identified deficiencies in the future development plans of the information systems. After analyzing and evaluating previous researches, 28 entities and 423 attributes were identified and the last entity-relationship model was created. Based on the presented model, a measuring tool was provided as a checklist so that banks can use it to measure their status in terms of the richness of existing data and to measure their readiness from the perspective of the data to do the analysis. To confirm the last data model, were used idea of ten experts by questionnaires and interviews in different sections such as customers and public banking, finance and support, e-banking, credit and corporate affairs, IT domain and international affairs in the bank. Also, using data collected from the researches were presented frequency diagrams of the algorithms, techniques, sampling methods, performance indexes and data mining soft­wares that used in the researches. To decide which data mining algorithms are most used in different domains as an example. Manuscript profile
      • Open Access Article

        6 - The study of the accuracy of real estate experts' evaluations using a data mining model (Case study of Mellat Bank)
        fatemeh davar
        As the main part of the financial system, banks always face different risks, the most important of which is the credit scoring risk and property valuation. One of the issues faced by property valuation experts is how to evaluate property prices. In general, court expert More
        As the main part of the financial system, banks always face different risks, the most important of which is the credit scoring risk and property valuation. One of the issues faced by property valuation experts is how to evaluate property prices. In general, court experts assess real estate based on price indices. In this research, the researcher aimed to verify the accuracy of valuation experts by using data mining models. This action has been taken to help bank managers and audit reporters to make better decisions about experts and their valuations. Using property valuation indexes and data mining, a predictive model has been developed to predict property prices, and a combination of FCM and K-NN algorithms has been used to achieve a high performance prediction model. This measure was able to greatly increase the predictive accuracy and increase the efficiency of the proposed model. The accuracy level in predicting valuated prices was 84.21% and the RMSE rate in its forecast was 0.43. The proposed approach was tested on real estate valuation data of the Mellat Bank. Manuscript profile
      • Open Access Article

        7 - Analyzing the impact of macroeconomic variables on customer churn banking industry With data mining approach
        Mehrnaz Motahari nia
        Today, customer knowledge and understanding of its needs have become a business imperative. Organizations need customer satisfaction to sustain their business and succeed in a competitive market. Knowing customers through customer behavior analysis is possible with the More
        Today, customer knowledge and understanding of its needs have become a business imperative. Organizations need customer satisfaction to sustain their business and succeed in a competitive market. Knowing customers through customer behavior analysis is possible with the use of new technologies such as data mining techniques for organizations. The purpose of this research is to investigate the effective factors on Customers churn in the banking industry. For this purpose, the transaction data of sales terminals of a payment service provider company (PSP) in Iran has been analyzed. In the proposed model using the WRFM method and combining it with the K-Means clustering algorithm, sales terminals are split and loyalty each month. Then, using the additive selection method plus L take R and the multivariate linear regression algorithm, the effective features The percentage of customers discarded is selected from the monthly economic indicators per month. Based on the results of the implementation of the three variables, the index of stock market value, inflation and the price of all coins are the most effective variables among the economic indicators under study. Manuscript profile