Association Rules In Data Mining Research And Of Teaching Quality Assessment

With the rapid increasing of data quantity in recent years, Data Mining has become a hotspot in many research fields such as Artificial Intelligence and Pattern Recognition, has received more and more attention. Mining association rules is one of the most active directions in data mining. This thesis studies and analyses the data mining techniques systematically, especially in mining association rule and its application for evaluating quality of teaching. The main contents are listed as follows:1. Design and analysis of the improved algorithm for Apriori. The existed documents of association rules, especially the classical Apriori algorithm and its many optimized techniques, are analyzed and studied widely. Based on them, we present an improved algorithm named Apriori-B. The algorithm mainly takes account of the generation bottleneck of frequent itemsets; it improves the Apriori algorithm by reducing scan times of transaction database etc. Successively the process of mining association rules using the Apriori-B is illustrated with an example. The performance comparison between the Apriori-B algorithm and Apriori algorithm is experimented. The results of the experiments are analyzed. All of the experiments reveal good performance of the improved algorithm.2. Research on the validity of association rules. In order to produce true and valid association rules, the present judgment criteria are association rules' support and association rules' confidence. If the association rules are generated according to these criteria, many of them are false or invalid. In order to reduce invalid association rules, we presented three improved methods. We define interestingness, validity and IAD (integrated appraisal degree) for the quality of association rule, and then add them into the judgment criteria. According to the value of them, association rules are classified into positive, invalid and negative association rules. In general, only positive association rules are valid, which is only a proportion of all strong association rules. Finally, an association rules mining algorithm based on new judgment criteria is presented and tested. The experiments show that the methods can obviously reduce invalid association rules.3. Application of mining association rules in teaching quality evaluation system. Using teachers' personnel database and part of the teaching evaluation data by investigation of students in Hunan City University, we produce large itemsets by the Apriori-B algorithm and mine association rules basing on new judgment criteria. The excavated critical factors affecting quality of teaching can provide the decision support information for the teaching administrative branch, which help us to carry out the teaching task much better and to enhance the quality of teaching.