Development and implementations of fuzzy decision tree algorithms

In this work, fundamentally fuzzy ID3 algorithm and the effects of T-operators on its reasoning procedure on numeric data is investigated and two novel approaches working on linguistic data have been proposed. Firstly, the information about fuzzy ID3 approach is presented. The usage of T-operators is given in details. The implementation is applied on geographic classification of virgin olive oil. Then, the first proposed approach on linguistic data is the FID3-L-WABL (Fuzzy ID3 Algorithm Based on Linguistic Data by Using WABL Defuzzification Method) which is a novel version of the known Fuzzy Interactive Dichotomizer 3 (Fuzzy ID3) classification algorithm working on linguistic data. This approach is performed on L-R (Left-Right) Fuzzy Number. Each fuzzy number is defuzzified by using WABL (Weighted Averaging Based on Levels). Then, Fuzzy c-means algorithm is adapted to obtain membership values of each fuzzy term for each fuzzy variable. Consequently, Fuzzy ID3 algorithm is applied. The second proposed approach is the FID3-LR (Fuzzy ID3 Algorithm for L-R Fuzzy Data) which is a mixture of FkM-F (Fuzzy k-means Clustering Model for Fuzzy data) clustering algorithm working on L-R fuzzy data and Fuzzy Interactive Dichotomizer 3 (Fuzzy ID3) classification algorithms. Fuzzy c-means algorithm was performed in MATLAB 2014a. The codes for the experiments, FuzzyID3 by using T-operators, FID3-L-WABL, and FID3-LR, have been developed in the MS Visual Studio C# IDE for the experimental study (Intel i7, 2.4 GHz, 4 Gb RAM). They have been designed as an integrated software system called as Fuzzy Artemis. In addition, OliveDeSoft is designed for current and future studies in order to analyze the olive oil quality and geographic characterization.