Quantitative Association Rules Mining Method Based On Trapezium-Cloud Model

Mining association rules is a main field in data mining . This thesis focus on the association rules mining. First, I studied the origin , classification background, idea, algorithms and current situation of association rules home and aboard, found out the advantages and disadvantages of it. Introduced two popular mining algorithms for boolean data?Apriori (produce large item set) and FP-tree(don' t produce large item set ), the advantages and disadvantages of them were compared. At present the algorithms to mine bool data is mature, but the algorithms to mine quantitative data is in underdevelopment. There so many methods to mine quantitative data, the easy way is to transform the quantitative data into boolean data, then mine them with the method the boolean data use it , but there are so many shortcomings of the traditional range-partition method that we can find few meaningful rules from the mining result.This thesis give a trapezium-cloud model based on the cloud model that academician-li brought forward. This model can perform qualitative and quantitative conversion and integrate fuzziness and randomness, so it can finish the natural conversion from quantitative data to qualitative concept. The concept we get from concept partition algorithms based on the trapezium-cloud model can reflect the real distributing of data, and the boundary of the concept is fuzzy and uncertain, so it is not parted forcedly, the rusult ocnform to human thinking. The meaning of the concept is reasonable. The rules mined by the quantitative association rules algorithms based on the trapezium梒loud model is effective and understandable, so we can find out more meaningful knowledge from the data.