Research And Realization Of Data Mining For University Asset

Data mining has attracted wide attention in information industry. University assets and analysis of data mining research is to make full use of the potential value of the asset data, which can provide scientific basis for both better management in universities and the decision-making.Association rule mining, as an important research field as well as one of the main research directions, has broad application value. It refers to the discovery of the links among items in a large number of transaction records, which indicate some association rule among database or data warehouse. People used to put forward many algorithms and its variants for mining association rules, among which the most famous is the Apriori algorithm. In association rule mining, frequent I/O operation will surely affect the efficiency of mining association rules. The main method to reduce I/O operation is to reduce the times of scanning data sets, to reduce the number of candidate itemsets, needed for counting the support degree, and to make the number of candidate itemsets close to that of frequent itemsets.Among the existing algorithms of mining association rules, the evaluation criteria of support and confidence is widely accepted. However some past applications show, data mining may lead to a large number of rules, yet most of which may be uninteresting or useless to the users, and may even be misleading. In order to solve this problem, this paper proposes an increased interest in threshold. When people mine an association rule, the most meaningful patterns can be achieved only when the degree of support, confidence and interest threshold are all greater than the minimum degree of them,This paper analyzes the current situation and characteristics in university assets data resource, builds data warehouse snowflake models and adopts the department constraints on data warehouse record for processing. Based on the research of association rule mining algorithm, and due to the classical Apriori algorithm’s disadvantages, such as too many times of scanning database system, heavy I/O load and producing a large number of irrelevant intermediate itemsets, we design a structured query language (SQL) operation system, and we also introduce interest as a criterion for the evaluation of effective association rule mining algorithm, which is to develop the rules interested by the user. The algorithm can reduce the candidate itemsets by adding users’interested items. As a result, it breaks the traditional algorithm steps, reduces the times of scanning the database, reduces the system I/O load, builds users’interest models, increases the readability of strong association rules resulting from algorithm, and improves the efficiency of algorithm. Experiment results show, making use of the sector constraints and the improved Apriori algorithm can effectively improve the mining speed and efficiency, and also dig out the association rules among the assets management indicators. Finally the data mining model is evaluated and new assets policy recommendations are suggested. Good results are achieved in the practical application and the level of scientific and practical management of assets in colleges and universities has been improved.