توسعه فازي الگوريتم Apriori براي كاوش قوانين وابستگي

چكيده انگليسي :

Fuzzy Extension of Apriori algorithm for Mining Association Rules Zohreh Ahmadi z ahmadidastjerdi@ec iut ac ir Date of Submission 2011 04 25 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language FarsiSupervisor Mohammad Hossein Saraee saraee@cc iut ac irAbstractIn the current world information is one of the most important producing factors So trying to extractinformation from data is one of the challenges in the information industry and the related area Volume ofdata is growing rapidly in all environments and in different ways It shows the complexity of changing datato information Data mining is one of the recent progresses in the field of data management In the datamining database theories artificial intelligence machine learning and statistics are combined to prepare anapplied area Data mining is composed of different methods which one of the most important of them isassociation rules mining The most currently used technique in association rules mining is Apriorialgorithm Various studies have been done already to develop association rules mining which the Apriorialgorithm has been a base for them The focus of this research is on a new sight to the association rulesmining according to the fuzzy logic After proposing fuzzy logic it is used in the intelligence systems because of its similarity to the human reasoning and then it is entered to the data mining rapidly In thecompare with classical logic fuzzy logic is a new ideology which is more adaptable to the necessities of thecurrent complex world In the fuzzy theory the truthness of expressions is a value between zero and one butin the classical logic the expressions are true or false So the results of applying fuzzy logic on differentproblems are more applicable and real In this research in addition to the review of recent previous studiesand considering their weaknesses and strengths a new algorithm named FTARM has been developed tofind fuzzy association rules from various quantitative data by considering a multi level fuzzy taxonomy anddifferent minimum supports for different items One of the most important applied ideas in this algorithm isits ability to apply on various datasets with different attributes including discrete and continuous values The ability of the proposed algorithm in considering fuzzy taxonomy and mining association rules in all ofits levels with all of the above assumptions is one of the most applicable strengths of it Also in this studya method is proposed to obtain fuzzy taxonomy based on a multi variate statistical analysis method namedas factor analysis Applying this method to extract the fuzzy taxonomy is suggested in the cases which thetaxonomy does not exist or its generating is difficult So using the factor analysis it is possible to extractthe fuzzy taxonomy The factor analysis method produces the hidden factors in the higher level byclassifying the attributes of the lower level The above methods have been implemented on two examples toshow the performance of the proposed algorithm The first example is in the field of aerology and includesaverage monthly temperature of 26 capital cities of Iran s province from 1991 to 2000 The second exampleis about the road transportation in Iran and the data of 14 atribbutes of the safty in the field of roadtransportation in 2008 have deen used Keywords Fuzzy association rules minimum support fuzzy taxonomy itemset

احمدي دستجردي، زهره

توسعه فازي الگوريتم Apriori براي كاوش قوانين وابستگي

كمترين پشتيبان , طبقه بندي فازي , مجموعه اقلام