توسعه مدل هاي دسته بندي و استخراج قواعد بر مبناي نوع داده

چكيده انگليسي :

Development of classification models and extracting rules based on data type Mohammad Taghi Rezvan taghi rezvan@in iut ac ir Date of defense 2014 11 08 Department of industrial engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree Ph D Language Farsi Supervisor Ali Zeinal Hamadani Hamadani@cc iut ac irAbstract Classification is one of the most common tasks of data mining and knowledge discovery whichmaps each item of the selected data onto one of a given set of classes Classification has countlessapplications in many fields including financial insurance medical social biological sciences etc Improving performance and capabilities have always attracted attention in this field Feature selection is a preprocessing procedure in pattern recognition and data mining Thisthesis uses rough set theory as an eff ective feature selection method A tree of the subsets of the original features set is developedand searched minimally to prune branches based on a monotonic property Starting the search froma greedy solution yields an effective and exact feature selection algorithm in rough set forcategorical datasets The capability of the algorithm is compared with full search Furthermore itssolution and computation time are compared with a meta heuristic algorithm The strengths and theweaknesses are described The classification models developed in this thesis are able to treatdifferent types of features such as numerical categorical and mixed features differently withouttransforming them In fact the distance or similarity measures of case based reasoning model arebuilt These measures consider the weight for each feature and handle categorical and numericalfeatures differently The proposed distance measures use the Euclidean distance for numericalfeatures and co occurrence of values for categorical features The proportional distribution ofdifferent categorical values of features is computed only with respect to the values of class featuresat two states without with considering the class of the cases The proposed case based reasoningmodels are implemented on categorical and mixed datasets and their performance is evaluated incomparison with the well known tools of classification Additionally a genetic algorithm similarto the introduced case based reasoning models is proposed which does not require thepreprocessing operations on features The proposed algorithm has suitable performance and canelicit the classification rules This algorithm is evaluated on categorical numerical and mixeddatasets and the obtained results are compared with other effective tools based on averageaccuracy rule set size and the length of the extracted rules The problem of sticker defect on cold rolling coils of Mobarakeh Steel Complex as aclassification problem is investigated to fulfill the practical perspective of thesis For this purpose the features which were effective in producing defect are determined from research and expertviewpoints and the available data are collected After refining the dataset and performing initialanalysis the performance of the proposed classifiers and some of the other well known methodsare used on datasets Accordingly the important features responsible for sticker defect areidentified Followed by the extraction of high accuracy classification rules used for setting differentprocess parameters so as to reduce or possibly omit sticker defect Keywords Classification Case based reseaning Rough set theory Genetic algorithm sticker defect

رضوان، محمدتقي

توسعه مدل هاي دسته بندي و استخراج قواعد بر مبناي نوع داده

استنتاج مبتني بر نمونه , نظريه مجموعه سخت , الگوريتم ژنتيك , عيب چسبندگي