Performance Analysis of Machine Learning Techniques and Fuzzy Rule Based Systems on Classification Problems

Abstract

Machine learning techniques and fuzzy rule-based systems (FRBS) are important tools to solve classification problems. In this paper, Machine learning and FRBS are used to design and implement systems for classification problems. The system is modelled using logistic regression (LR), random forest classifier (RF), gradient boosting classifier, gaussian naive bayes (NB), decision tree classifier (DT), K-nearest neighbour classifier and support vector machine (SVM). The prediction accuracy of these machine learning models is approximately 77%. Also, the same classification problem was modeled in FRBS and a big improvement was achieved in the accuracy, which was 99.7% with fuzzy partition Count 5. Hence, in this work it is concluded that FRBS are more accurate than machine learning and prediction Models.

1991 Mathematics Subject Classification: 03B52.

Corresponding Author: Piyush Kumar Tripathi