Research And Application Of Association Rule Mining Based On Geographical Information

With the development of the information industry, the data produced in society has shown exponential growth. This large amount of data contains priceless information for people to explore. At this time, a new technology, Knowledge Discovery in Database, which is aimed at finding valuable information in large amount of data through effective methods is born. Today, the data mining algorithm can be applied to plenty of fields Because of the development of the algorithm, people have been paid more and more attention to the research and application of geographic information since the development of the search engine. With the boom of the mobile Internet, more and more research of the data mining in the geographic information and its related information has been conducted.In this paper, the landmark data surrounding North 4th Ring Road and South 4th Ring Road within the range of 10 kilometers is selected from the Beijing open GIS data set. According to the high prices and intensive population of North 4th Ring Road, the mining of geographic information rules near North 4th Ring Road is conducted to reveal the rules of the influence on the prices and prosperity of the area while comparing different types of the constructions. A reasonable and scientific explanation is made based on the results.In consideration of the specificity of geographic position information applied in the experiment, the pretreatment of the mining data and the improvement of mining process are suggested to achieve the generation and interpretation of the correct mining results in this paper.(1) To transfer the relational database into the boolean database, the clustering algorithm is necessary. The density based clustering algorithm, which is applied to cluster the geographic information collection as transactions, is applied to solve the problem that the geographical location information cannot be the input of the Apriori algorithm process in the boolean database because of its discrete type.(2) The transaction classification method of the association rules mining algorithm is improved due to the large capacity of large database, low efficiency of the mining process and the large consuming of time. The improvement of transaction classification combined with FP-tree algorithm is proposed to accommodate the different scale of data by split the dataset. The results show that the improvement enhance both the efficiency and the capacity.(3) The extended information is introduced into the mining process. The geographic distance information is added into the mining process to generate more specific results so that further explanation can be made.