As the credit industry has been growing rapidly, huge number of consumers’ credit data are collected by the credit department of the bank and credit scoring has become a very important issue. Usually, a large amount of redundant information and features are involved in the credit dataset, which leads to lower accuracy and higher complexity of the credit scoring model, so, effective feature selection methods are necessary for credit dataset with huge number of features. This paper aims at comparing seven well-known feature selection methods for credit scoring. Which are t-test, principle component analysis (PCA), factor analysis (FA), stepwise regression, Rough Set (RS), Classification and regression tree (CART) and Multivariate adaptive regression splines (MARS). Support vector machine (SVM) is used as the classification model. Two credit scoring databases are used in order to provide a reliable conclusion. Regarding the experimental results, the CART and MARS methods outperform the other methods by the overall accuracy and type I error and type II error.
{"title":"Feature Selection Based on SVM for Credit Scoring","authors":"Ping Yao","doi":"10.1109/CINC.2009.36","DOIUrl":"https://doi.org/10.1109/CINC.2009.36","url":null,"abstract":"As the credit industry has been growing rapidly, huge number of consumers’ credit data are collected by the credit department of the bank and credit scoring has become a very important issue. Usually, a large amount of redundant information and features are involved in the credit dataset, which leads to lower accuracy and higher complexity of the credit scoring model, so, effective feature selection methods are necessary for credit dataset with huge number of features. This paper aims at comparing seven well-known feature selection methods for credit scoring. Which are t-test, principle component analysis (PCA), factor analysis (FA), stepwise regression, Rough Set (RS), Classification and regression tree (CART) and Multivariate adaptive regression splines (MARS). Support vector machine (SVM) is used as the classification model. Two credit scoring databases are used in order to provide a reliable conclusion. Regarding the experimental results, the CART and MARS methods outperform the other methods by the overall accuracy and type I error and type II error.","PeriodicalId":173506,"journal":{"name":"2009 International Conference on Computational Intelligence and Natural Computing","volume":"155 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115153831","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
First, good performance of Bayesian variable selection (BVS for short) in a variety of applications is introduced. Then, we will give a theoretical explanation why BVS works so well in linear models. We assume the true regression coefficients vector of the linear model is sparsity, in a sense that some regression coefficients are bounded from zero while the rest are exactly zero. In this case, under some conditions, BVS will show it can select the true model by means of giving a consistent estimate of the true regression coefficients vector.
{"title":"On the Consistency of Bayesian Variable Selection for High Dimensional Linear Models","authors":"Shuyun Wang, Y. Luan","doi":"10.1109/CINC.2009.189","DOIUrl":"https://doi.org/10.1109/CINC.2009.189","url":null,"abstract":"First, good performance of Bayesian variable selection (BVS for short) in a variety of applications is introduced. Then, we will give a theoretical explanation why BVS works so well in linear models. We assume the true regression coefficients vector of the linear model is sparsity, in a sense that some regression coefficients are bounded from zero while the rest are exactly zero. In this case, under some conditions, BVS will show it can select the true model by means of giving a consistent estimate of the true regression coefficients vector.","PeriodicalId":173506,"journal":{"name":"2009 International Conference on Computational Intelligence and Natural Computing","volume":"56 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124275882","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Shaowei Su, Zhentan Feng, Jinlin Wang, Yifeng Lu, Jiali You
Data scheduling is the key issue in peer-to-peer streaming system, especially in peer-to-peer VoD system. This paper mainly focuses on data chunks’ priority definition of data scheduling in peer-to-peer VoD system. After carefully studying the meanings of regular rarest first scheduling and random scheduling, a Local-Rarest-Random-Heurist (LR2H) scheduling is proposed in order to fully use the resources of strong peers in the system. LR2H fully considers the meanings of rarest first priority and random priority. It brings abundant copies of data chunks in the system by using part of rarest priority and it also avoids the concentrate request in some data chunks which have same rarest priority by adding a random jitter, and then maximize total priority of every scheduling by using heuristic algorithm. Simulation proves that LR2H has achieved a scheduling effect about 30% better than normal rarest first or random scheduling.
{"title":"A Local Rarest-Random-Heuristic Data Scheduling for P2P VoD System","authors":"Shaowei Su, Zhentan Feng, Jinlin Wang, Yifeng Lu, Jiali You","doi":"10.1109/CINC.2009.21","DOIUrl":"https://doi.org/10.1109/CINC.2009.21","url":null,"abstract":"Data scheduling is the key issue in peer-to-peer streaming system, especially in peer-to-peer VoD system. This paper mainly focuses on data chunks’ priority definition of data scheduling in peer-to-peer VoD system. After carefully studying the meanings of regular rarest first scheduling and random scheduling, a Local-Rarest-Random-Heurist (LR2H) scheduling is proposed in order to fully use the resources of strong peers in the system. LR2H fully considers the meanings of rarest first priority and random priority. It brings abundant copies of data chunks in the system by using part of rarest priority and it also avoids the concentrate request in some data chunks which have same rarest priority by adding a random jitter, and then maximize total priority of every scheduling by using heuristic algorithm. Simulation proves that LR2H has achieved a scheduling effect about 30% better than normal rarest first or random scheduling.","PeriodicalId":173506,"journal":{"name":"2009 International Conference on Computational Intelligence and Natural Computing","volume":"164 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121358976","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper, we analyze the Hang Seng Index data for the 22-year period, from December 31, 1986, to June 6,2008 in the Hongkong stock market, a total of 5315 trading days. Using rescaled range method, we study how the threshold value q affects the correlations of the return intervals s r(τ ) between events above a certain threshold q. We find that: i) both return intervals obtained by different threshold q and the original series are arranged in long-range dependence behavior; ii) the correlations of the return intervals grow stronger when the threshold q is larger.
{"title":"Return Intervals Analysis of the Hong Kong Stock Market","authors":"Hong Zhang, Nianpeng Wang, Keqiang Dong","doi":"10.1109/CINC.2009.108","DOIUrl":"https://doi.org/10.1109/CINC.2009.108","url":null,"abstract":"In this paper, we analyze the Hang Seng Index data for the 22-year period, from December 31, 1986, to June 6,2008 in the Hongkong stock market, a total of 5315 trading days. Using rescaled range method, we study how the threshold value q affects the correlations of the return intervals s r(τ ) between events above a certain threshold q. We find that: i) both return intervals obtained by different threshold q and the original series are arranged in long-range dependence behavior; ii) the correlations of the return intervals grow stronger when the threshold q is larger.","PeriodicalId":173506,"journal":{"name":"2009 International Conference on Computational Intelligence and Natural Computing","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124821684","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper describes a framework for the design of a test tool that could generate test cases automatically based on given BPEL specifications. The key problems that need to be addressed are how to transform the BPEL specifications into a HPN, and how to design a script language to describe the test case generation that according to the characteristics of BPEL. A BPEL Specification Analyzer and a Test Script Language are presented. A tool called BPEL-based Testing Automatic has been designed and partially implemented. BTA will take a user-defined test case template and the set of test data generated to produce the executable test cases.
{"title":"Test Case Generation Method for BPEL-Based Testing","authors":"Wenli Dong","doi":"10.1109/CINC.2009.229","DOIUrl":"https://doi.org/10.1109/CINC.2009.229","url":null,"abstract":"This paper describes a framework for the design of a test tool that could generate test cases automatically based on given BPEL specifications. The key problems that need to be addressed are how to transform the BPEL specifications into a HPN, and how to design a script language to describe the test case generation that according to the characteristics of BPEL. A BPEL Specification Analyzer and a Test Script Language are presented. A tool called BPEL-based Testing Automatic has been designed and partially implemented. BTA will take a user-defined test case template and the set of test data generated to produce the executable test cases.","PeriodicalId":173506,"journal":{"name":"2009 International Conference on Computational Intelligence and Natural Computing","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125850428","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Hongli Cai, Zhang Yang, Xianlin Zhou, Peng Jing, Jianliang Wang
Aspect-oriented programming may improve the design level of software, the reusability of components and the implementation of separation of concerns. Component-based software development approach is one of the most promising solutions for the emerging high development cost, low productivity, unmanageable software equality and high risk. This approach, however, encounters the separation of concerns that is easy to lead to the code-tangling and code-scattering. This paper aims to solve this problem on requirement level through the aspect-oriented requirement. At the same time, we also give concerns to the reuse of aspect.
{"title":"Aspect-oriented Requirement and Reuse Aspect","authors":"Hongli Cai, Zhang Yang, Xianlin Zhou, Peng Jing, Jianliang Wang","doi":"10.1109/CINC.2009.172","DOIUrl":"https://doi.org/10.1109/CINC.2009.172","url":null,"abstract":"Aspect-oriented programming may improve the design level of software, the reusability of components and the implementation of separation of concerns. Component-based software development approach is one of the most promising solutions for the emerging high development cost, low productivity, unmanageable software equality and high risk. This approach, however, encounters the separation of concerns that is easy to lead to the code-tangling and code-scattering. This paper aims to solve this problem on requirement level through the aspect-oriented requirement. At the same time, we also give concerns to the reuse of aspect.","PeriodicalId":173506,"journal":{"name":"2009 International Conference on Computational Intelligence and Natural Computing","volume":"945 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123302040","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
As an important task of multi-relational data mining, multi-relational classification can directly look for patterns that involve multiple relations from a relational database and have more advantages than propositional data mining approaches. According to the differences in knowledge representation and strategy, the paper researched three kind of multi-relational classification approaches that are ILP based, graph-based and relational database-based classification approaches and discussed each relational classification technology, their characteristics, the comparisons and several challenging researching problems in detail.
{"title":"Research on Multi-Relational Classification Approaches","authors":"Peng Zhen, Lifeng Wu, Xiaoju Wang","doi":"10.1109/CINC.2009.166","DOIUrl":"https://doi.org/10.1109/CINC.2009.166","url":null,"abstract":"As an important task of multi-relational data mining, multi-relational classification can directly look for patterns that involve multiple relations from a relational database and have more advantages than propositional data mining approaches. According to the differences in knowledge representation and strategy, the paper researched three kind of multi-relational classification approaches that are ILP based, graph-based and relational database-based classification approaches and discussed each relational classification technology, their characteristics, the comparisons and several challenging researching problems in detail.","PeriodicalId":173506,"journal":{"name":"2009 International Conference on Computational Intelligence and Natural Computing","volume":"102 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114893936","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
It is well known that it plays an important role to determine threshold pressure gradient (TPG) in developing the low permeability oil field, and it directly influences the accuracy of reservoir pressure and developing amount. Threshold pressure gradient becomes nonlinear relation with such factors that may influence accuracy as permeability, viscosity and density of fluid and porosity and so on. Such a problem of nonlinear nature can be solved by RBF neural network systems. Based on above thought, authors of this paper predict the TPG using RBF neural network. This approach has further been tested and verified by actual determining results .The experimental results show that RBF neural network is an effective method for TPG prediction with good precision. The application of this approach can supply basic data for developing oil field so as to save cost and labor
{"title":"Application of RBF Algorithm in Prediction of Threshold Pressure Gradient","authors":"Chang-jun Zhu, Xiujuan Zhao, Wei-hua Yang","doi":"10.1109/CINC.2009.138","DOIUrl":"https://doi.org/10.1109/CINC.2009.138","url":null,"abstract":"It is well known that it plays an important role to determine threshold pressure gradient (TPG) in developing the low permeability oil field, and it directly influences the accuracy of reservoir pressure and developing amount. Threshold pressure gradient becomes nonlinear relation with such factors that may influence accuracy as permeability, viscosity and density of fluid and porosity and so on. Such a problem of nonlinear nature can be solved by RBF neural network systems. Based on above thought, authors of this paper predict the TPG using RBF neural network. This approach has further been tested and verified by actual determining results .The experimental results show that RBF neural network is an effective method for TPG prediction with good precision. The application of this approach can supply basic data for developing oil field so as to save cost and labor","PeriodicalId":173506,"journal":{"name":"2009 International Conference on Computational Intelligence and Natural Computing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116235285","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With reference to the idea of component design for workshop production business, This paper puts forward a Reconfigurable Manufacturing Execution System (RMES)based on workflow and Multi-Agent, establish it’s function architecture and operation control mechanism. Under the architecture of RMES, the configurable workshop production business can be handled, and the MES application system can correspondingly adjust with its change to realize the personalized production management system. By the use of software component, RMES can increase the adaptability and reconstructivity of MES effectively.
{"title":"A Reconfigurable Manufacturing Execution System and its Component Reuse","authors":"L. Zhaohui, Chen Yan, C. Xiuquan","doi":"10.1109/CINC.2009.259","DOIUrl":"https://doi.org/10.1109/CINC.2009.259","url":null,"abstract":"With reference to the idea of component design for workshop production business, This paper puts forward a Reconfigurable Manufacturing Execution System (RMES)based on workflow and Multi-Agent, establish it’s function architecture and operation control mechanism. Under the architecture of RMES, the configurable workshop production business can be handled, and the MES application system can correspondingly adjust with its change to realize the personalized production management system. By the use of software component, RMES can increase the adaptability and reconstructivity of MES effectively.","PeriodicalId":173506,"journal":{"name":"2009 International Conference on Computational Intelligence and Natural Computing","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122311462","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Wavelet packet decomposition not only has the decompose effect at low-frequency by using wavelet decomposition, but also has the decompose effect at high-frequency where can not do by using wavelet decomposition. In this paper, the wavelet packet decomposition algorithm was proposed and applied to glass-image recognition. Compared with other feature extracting technologies such as Zernike’s moments and wavelet transformation, the experiments proved that the wavelet packet decomposition was the best on both precision and efficiency
{"title":"An Algorithm of Glass-Image Recognition Based on Wavelet Packet Decomposition","authors":"Huan Liang, W. Zhihua","doi":"10.1109/CINC.2009.29","DOIUrl":"https://doi.org/10.1109/CINC.2009.29","url":null,"abstract":"Wavelet packet decomposition not only has the decompose effect at low-frequency by using wavelet decomposition, but also has the decompose effect at high-frequency where can not do by using wavelet decomposition. In this paper, the wavelet packet decomposition algorithm was proposed and applied to glass-image recognition. Compared with other feature extracting technologies such as Zernike’s moments and wavelet transformation, the experiments proved that the wavelet packet decomposition was the best on both precision and efficiency","PeriodicalId":173506,"journal":{"name":"2009 International Conference on Computational Intelligence and Natural Computing","volume":"80 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122325845","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：