A hybrid decision tree/genetic algorithm method for data mining

Abstract

This paper addresses the well-known classification task of data mining, where the objective is to predict the class which an example belongs to. Discovered knowledge is expressed in the form of high-level, easy-to-interpret classification rules. In order to discover classification rules, we propose a hybrid decision tree/genetic algorithm method. The central idea of this hybrid method involves the concept of small disjuncts in data mining, as follows. In essence, a set of classification rules can be regarded as a logical disjunction of rules, so that each rule can be regarded as a disjunct. A small disjunct is a rule covering a small number of examples. Due to their nature, small disjuncts are error prone. However, although each small disjunct covers just a few examples, the set of all small disjuncts can cover a large number of examples, so that it is important to develop new approaches to cope with the problem of small disjuncts. In our hybrid approach, we have developed two genetic algorithms (GA) specifically designed for discovering rules covering examples belonging to small disjuncts, whereas a conventional decision tree algorithm is used to produce rules covering examples belonging to large disjuncts. We present results evaluating the performance of the hybrid method in 22 real-world data sets. _____________________________________________________________________________________________

A multi-objective genetic optimization of interpretability-oriented fuzzy rule-based classifiers

“…[79] Table 6 concerns results from [79], where 10 different methods of automatic classifier construction based on data were considered. We chose 7 out of them for comparative analysis: C45R (C45Rules technique of generating a set M a n u s c r i p t of production rules from a tree produced by C4.5 algorithm) [83], RIPPER (improvement of the efficient Incremental Reduced Error Pruning algorithm) [84], MPLCS (a Memetic Pittsburgh Learning Classifier System) [85], AntMin+ (Ant Colony-based data miner to extract classification rules inspired by the research on the behaviour of real ant colonies) [86], CORE (COevolutionary Rule Extractor) [87], DTGA (hybrid decision tree/genetic algorithm discovering rules on small disjuncts) [88], and GFS-GP (Genetic Programming algorithm used to learn fuzzy rule-based classifiers) [89]. Three remaining methods of [79] were skipped: ILGA (Incremental Learning approach to Genetic Algorithms) [90], SLAVE (Structural Learning Algorithm on Vague Environment) [91], and the ICRM (Interpretable Classification Rule Mining algorithm) method proposed by the authors of [79].…”

Fuzzy dual experience-based design evaluation model for integrating engineering design into customer responses

“…There are many studies trying to improve classifying efficiency. Chang and Liu [11] and Carvalho and Freitas [8] applied decision tree for other method such as K-mean, fuzzy C mean. Chandra and Varghese [10] proposed a binary decision tree algorithm using the Gini index as a split measure.…”

“…Total cost (TC) is by defining and discussing the links between R&D cost, market capital, and design quality for market share in Eq. (8). Market share can be gained by attracting customers with preferences more distant from the target market.…”

Developing Best Practices for Descriptor‐Based Property Prediction: Appropriate Matching of Datasets, Descriptors, Methods, and Expectations

“…The most common strategy is to evaluate every possible subtree, working from the leaves backward, for possible replacement by a terminal node. Nontrivial heuristics including genetic algorithms [114] have also been demonstrated as useful pruning strategies [115,116].…”