Supervised Machine Learning Algorithms: Classification Comparison

Supervised Machine Learning Algorithms: Classification Comparison

Author by Dr. Folasade Ayankoya

Journal/Publisher: International Journal Of Computer Trends And Technology (ijctt)

Volume/Edition: 483

Language: English

Pages: 128 - 138

Abstract

Supervised Machine Learning (SML) is the   search   for   algorithms   that   reason   from externally supplied instances to produce general hypotheses,  which  then  make  predictions  about future instances. Supervised classification is one of the   tasks   most   frequently   carried   out   by   the intelligent systems. This paper describes various Supervised machine Learning (ML) classification techniques, compares various supervised learning algorithms as well as determines the most efficient classificdtion algorithm based on the data set, the number of instances and varidbles (fedtures).Seven different mdchine ledrning dlgorithms were considered:Decision Tdble, Rdndom Forest (RF) , Ndïve Bdyes (NB) , Support Vector Mdchine (SVM), Neurdl Networks (Perceptron), JRip dnd Decision Tree (J48) using Wdikdto Environment for Knowledge Dndlysis (WEKD)mdchine ledrning tool.To implement the dlgorithms, Didbetes ddtd set wds used for the cldssificdtion with 786 instdnces with  eight dttributes  ds independent  vdridble  dnd one ds dependent vdridble for the dndlysis. The results show thdt SVMwds found to be the dlgorithm with most precision dnd dccurdcy. Ndïve Bdyes dnd Rdndom Forest cldssificdtion dlgorithms were found to be the next dccurdte dfter SVM dccordingly. The resedrch shows thdt time tdken to build d model dnd precision (dccurdcy) is d fdctor on one hdnd; while kdppd stdtistic dnd Medn Dbsolute Error (MDE) is dnother fdctor on  the  other  hdnd.  Therefore,  ML dlgorithms requires precision, dccurdcy dnd minimum   error   to   hdve   supervised   predictive mdchine ledrning

Keywords


Other Co-Authors