Mining Based On Association Rules And Cluster Analysis Of Abnormal Weather

Data Mining or Knowledge Discovery arisen in the late 1980s has developed quickly, which based on statistics, artificial intelligence (especially in the field of machine learning) and database, etc. many sorts of technologies. It has become a hotspot in the fields of artificial intelligence and database technology. Association rule mining and clustering analysis mining are two primary and important sub-branches of data mining. They have been used widely in selective marketing,telecom business, finance management, electronic government, biology, medicine, industry and so on.Along with the wide use of computer technology, the informationalization level in the field of meteorology is on the up and up. A lot of data about weather has been gathered, which concealed valuable information. It has become an important task for researchers about how to utilize the data and obtain useful information hidden in them. The features and relations hidden in the data can be mined and gained by using different methods.This thesis briefly introduces the present situations about using data mining methods to mine meteorological data, the terms and definitions of weather and meteorological disasters. Then making a basic study to classic association rule algorithms and clustering algorithms, including concepts, types and mining steps of them. Mainly focusing on Apriori algorithm, k-means algorithm and k-medoids algorithm. According to the practice in data mining, offering a tri-dividing clustering algorithm around extreme values and giving detailed explanations and analysis. The algorithm originates from k-means algorithm and k-medoids algorithm and is improved to meet the needs of data mining greatly. In chapter 6, the term of times-proportion factor has been given and used to predict the future temperature and gets a good result. The tri-dividing clustering algorithm around extreme values and Apriori algorithm are used to mine the meteorological data in Chongqing city and the concrete mining process has been shown. A series of mining results are of great importance. They are very significant to discover the distribution features of disastrous weather and realated elements causing it. In the last chapter of this thesis the author looks into the splendid future on using data mining methods to mine meteorological data. At the same time, the flaws of the methods in predicting future weather are given.When data mining technology is used in the process of meteorological data collecting,processing and application to discover the distribution features of disastrous weather and realated elements causing it,measures against disastrous weather are taken swiftly by utilizing the former reults,better service can be provided for scientific decision.