利用資料場模型及模糊關聯法則挖掘空間資料中共同出現的關係

Mining Spatial Colocation Patterns Using Data Field Model and Fuzzy Association Rules

Martinus(Martinus)

指導教授 ： 蕭翰文

亞洲大學/資訊學院/資訊工程學系碩士班/碩士(2006年)

摘要

Scientists in many researches have been using computer technologies lately. GIS, GPS have been helping scientists in doing many kinds of researches. Geographical data as a result from GPS were available in electronic format. This type of data can be treated as a spatial data. And by using colocation pattern mining, we would discover associations between spatial features. The first thing we do was developed a data set generator. Data sets that are generated by data set generator then processed using the proposed approach.
The system we proposed was a two steps system. The first step was doing a segmentation to produce the transaction from the data set. The segmentation was using a fix threshold segmentation and the threshold was 3*sigma. Sigma in this study is our way to measure closeness of a point to its neighbors. Sigma is a distance value that will bring the entropy of the whole data set into it minimum, sigma was calculated using data field model. And the second one was doing a fuzzy association rule mining where we introduce the transaction into a fuzzy membership function. After fuzzfied the data set then we counted the fuzzy support values and fuzzy confidence values. The infrequent rules then pruned using an apriori-like algorithm. The result of the approach then come as these manners, feature a will be whether near or close or far from feature b.

關鍵字

colocation pattern mining, fuzzy association rule, data field model.

並列摘要

Scientists in many researches have been using computer technologies lately. GIS, GPS have been helping scientists in doing many kinds of researches. Geographical data as a result from GPS were available in electronic format. This type of data can be treated as a spatial data. And by using colocation pattern mining, we would discover associations between spatial features. The first thing we do was developed a data set generator. Data sets that are generated by data set generator then processed using the proposed approach.
The system we proposed was a two steps system. The first step was doing a segmentation to produce the transaction from the data set. The segmentation was using a fix threshold segmentation and the threshold was 3*sigma. Sigma in this study is our way to measure closeness of a point to its neighbors. Sigma is a distance value that will bring the entropy of the whole data set into it minimum, sigma was calculated using data field model. And the second one was doing a fuzzy association rule mining where we introduce the transaction into a fuzzy membership function. After fuzzfied the data set then we counted the fuzzy support values and fuzzy confidence values. The infrequent rules then pruned using an apriori-like algorithm. The result of the approach then come as these manners, feature a will be whether near or close or far from feature b.

並列關鍵字

colocation pattern mining, fuzzy association rule, data field model.

參考文獻

Hsiao, H. W., and Tsai, M. H., (2006) Spatial Data Mining of Colocation Patterns for Decision Support in Agriculture, Asian Journal of Health And Information Sciences

Agarwal, R., Imielinksi, T. and Swami, A. (1993) Mining association rules between sets of item in large database, The ACM SIGMOD Conference, Washington D.C., USA

Chawla, S., Shekhar, S., Wu, W., and Ozesmi, U., (2001) Modeling spatial dependencies for mining geospatial data: An introduction, Geographic Data Mining and Knowledge

Chris D., T., and Cameron, A., et. al. (2004) Extinction risk from climate change, Nature 427, pp145-148

Cover, T.M., and Thomas, J.A., (1991) Elements of Information Theory, John Wiley & Sons, New York, USA,