Mining Multilevel Spatial Association Rules And The Performance Evaluation Methods

Data mining is the process of mining the hidden and valuable knowledge from a large number of incomplete and random data sets under the existing technical conditions. With the rapid development of modern technology, the knowledge of modern society is developing rapidly in an unprecedented way,and the data mining is becoming more and more important. A new method, which based on the Apriori algorithm, has been designed in this paper. It improves the Apriori algorithm by combining the advantages of heapsort and linked list, put forward an improved Apriori algorithm and apply it to the mining multi-level spatial association rules. The performance of evaluation methods has also been analyzed here.Mining the Association rules is one of the most commonly used methods in data mining. The association rules is important to the data relationship mining in a dataset. In this paper:a new method of discovering association rules, which combines the characteristics of Apriori algorithm and the characteristics of heapsort has been proposed. Through the analysis of the experimental results, it has been found that the improved algorithm is effective. The obtained results, the more important association rules, are the first "N" items from highest frequency. The improved algorithm not only optimizes the selection of datasets and the calculation of generates, but also optimizes the pruning process. In addition, this new algorithm can use the existing data, in some extent to combine the existing results and the new data obtained in the future, to easily mine the important rules and infer corresponded rules..Finally, the improved algorithm has been applied to mine the multi-level spatial association rules, and the experimental results have been analyzed.The evaluation method of association rules is-to evaluate the algorithm of association rules from many aspects.The aim is to sufficiently compare the algorithm performance, The study has been carried out from two aspects, subjective and objective. In subjective study, the evaluation method which concerns about the comparison between human expectations and practical outcome has been analyzed. In objective study, the evaluation of the accuracy, space efficiency, running time, and the database accessing has been considered and analyzed. Then, the characteristics of data, which obtained from multi-level spatial mining process, have been briefly introduced in this paper. Finally, the multi-level spatial mined results have been compared with results obtained from other spatial based algorithms. Through the analysis of the experimental results, it has been found that the improved algorithm is effective, and achieves much better performance of extracting the spatial information in the top layer of spatial data.