Research Of An Improved Apriori Algorithm On Helping Recognize Poor College Students

Data mining is a process of extracting previously unknown,valid and actionable information from large database or data warehouse and then using information so derived to make crucial business and strategic decisions.Association rule mining remains a very popular and effective method to extract meaningful information from large datasets.It tries to find possible associations between items in large transaction based datasets.Apriori introduced by Agrawal in 1993 is a classic algorithm for learning association rules.The objective of this thesis is to improve the efficiency of Apriori algorithm on helping recognize poor college students.In order to make poor college students concentrate on their study without extra worries about their lives,China has many preferential policies to help them.It is important how to identify them through their consumption patterns,such as the consumption data of campus smart card.Most of the previous studies adopt Apriori algorithm and its variants.However,theses algorithms may still suffer from the following two nontrivial costs:(1)The candidate item set generated possibly is huge and results in combination explosion.(2)Frequently scanning database brings heavy load to disk I/O equipments,that is,after one candidate item set is generated;Apriori algorithm must scan database to determine whether each item is a frequent item or not.The thesis improves the efficiency of Apriori algorithm by using a partitioning-based method to avoid the costly generation of a large number of candidate sets and scan the database only once,It also discusses the implementation details of this improved algorithm and its application in helping recognize poor college students.The experiment makes use of the consumption data of campus smart card in 2015-2016 year in Guangdong University of Foreign Studies South China Business College.The improved algorithm is implemented through the MySQL community server 5.6 platform.Finally,the association rules mined from the consumption data of campus smart card can be used to predict the consumption behavior of poor college students.The result provides support for student-affairs-administrators identifying them.