Research Of Assocation Rules And Its Application To The Analysis Of Academic Achievements

With the development of information technology and wide application of database, the size of data that people faced are rapidly expansion. Facing these massive and chaos numbers of data, the traditional simple methods obviously can't meet people's needs on data processing. Therefore, seeking a technology to analyze large amounts of data has become an urgent demand. The emergence of data mining technology solves this problem exactly. Association rules is one of the most widely used and mature technology in data mining. It is mainly used for finding guiding rules in large and complex data.Association rules mining can be divided into two steps: mining frequent item sets and generate association rules. The first step is the key of the algorithm. Therefore, how to improve it's efficiency has become a priority issue in current research. This paper analyzes the characteristics and the lack of Apriori algorithm, and an imporoved algorithm AprioriTid. Using the idea of AprioriTid an improved algorithm is proposed which can make up the flaw of Apriori very well. The set of affair item are used to add up the count of candidate item sets. This paper use generating candidate item sets and pruning strategy which based on allocation index mechanism to reduce comparing number, and a quick connection service is also given for joining two lists. The experiment proved that the algorithm is efficient. To make up the flaw of association rule measures existed now, with introducing U-Testing, the effect as a new evaluation criterion is proposed. The new evaluation criterion can make up the flaw of the current measures effectively, and can identify invalid and negative association rules. The effect criterion divides positive association rules into weak and strong association rules, which can provide more effective rules to people.A university's computer science and software collage 2006 grade students'achievement of all the courses as the research object, using the standard data mining process and improved algorithms, this paper intends to find association rules between courses and course types, also give the mining results and explain some of them. At last, the framework and database of academic achievement analysis system are designed. The system realizes the functions of data selection, data cleaning and data conversion. At the beginning of data mining, according to different needs the users can choose data content. A rule template is introduced at the mining process and graph is implemented in this paper as for the visualization, which can help the users find interesting rules from large number of association rules quickly. Users can also sort the rules of mining and store the effective rules into the rule base. The rules can prompt early warning on student achievement and help teach manager arrangement course reasonably, so as to improve the quality of teaching in schools.