Accurate multi-criteria decision making methodology for recommending machine learning algorithm

详细信息    查看全文

• 作者：Rahman Ali^a ; ^b ; ^{rahmanac1@gmail.com} ; Sungyoung Lee^a ; ^{sylee@oslab.khu.ac.kr} ; Tae Choong Chung^a ; ^{tcchung@khu.ac.kr}
• 关键词：Multi-criteria decision making ; Algorithm recommendation ; Algorithm selection ; Classification algorithms ; Classifiers recommendation ; TOPSIS ; Ranking classifiers
• 刊名：Expert Systems with Applications
• 出版年：2017
• 出版时间：1 April 2017
• 年：2017
• 卷：71
• 期：Complete
• 页码：257-278
• 全文大小：1969 K
• 卷排序：71

文摘

A multi-criteria decision making methodology is proposed to select best classifier. NGT-based method is adapted to choose suitable classifier's evaluation metrics. Accuracy, time complexity and consistency based new ranking criteria is designed. AHP-based method is adapted to estimate relative weights for evaluation metrics. Classifiers are ranked based on relative closeness score, computed using TOPSIS.