| Summary: | Classification in data mining is very extensive research area. Decision trees have been found very effective for classification of huge and frequently modifiable databases e.g., Stock Exchange, Shopping Mall etc. We build a decision tree from a training set consists of two phases. In the first phase the initial Linguistic Rules-Tree (LR¬Tree) has been constructed. In LR-Tree we have combined fuzzy logics and decision tree. First we evaluate fuzzy membership function from training data for each attribute in class then apply our fuzzy linguistic approach which is associated with decision tree that provides for a fine grain description of classified items adequate for human reasoning. Consequently our approach will be able to handle training data with missing attribute values, handling attributes with differing costs, improving computational efficiency. But LR-Tree may not be the best generalization due to over-fitting so in the second phase, we will propose a novel frequent pattern mining tree called Compact Linguistic Rules-Tree (CLR-Tree) that remove some branches and nodes to improve the accuracy of the classifier. In this paper, we have concentrated on construction phase and hope that after completing the construction phase we will proof that the CLR- Treeis efficient and scalable for mining both long and short frequent patterns.
|
|---|