The Computation Trend of Fuzzy Rules for Effective Decision Support Mechanism on Basis of Supervised Learning for Multiple Periods

Singh B., Henge S. K.

4th International Conference on Intelligent and Fuzzy Systems (INFUS), Bornova, Turkey, 19 - 21 July 2022, vol.505, pp.117-123 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 505
  • Doi Number: 10.1007/978-3-031-09176-6_14
  • City: Bornova
  • Country: Turkey
  • Page Numbers: pp.117-123
  • Keywords: Neural fuzzy inference system (NFIS), Supervised learning (SL), K-nearest neighbors algorithm (KNA), Logistic regression (LR), Random forest classification (RFC), Naive Bayes Gaussian algorithms (NBGA), Risk management (RM)
  • Istanbul Technical University Affiliated: No


This paper elaborates the attempt to measure the performance of neural fuzzy inference systems along with five different machine-learning algorithms for a 10-year dataset. Several models developed with machine learning concepts were evaluating on a short-term dataset that imparted a very limited values of prediction accuracy. Background study reveals the utilization of fuzzy inference systems using multiple rules forming a complex structure for computation with lagging issues in performance in terms of executive time. This paper has composed with ML based algorithmic techniques along with neural network and fuzzy inference system for stock market prediction. The methods involve the integration of fuzzy rules for the decision making process on the basis of technical indicators.Consequently, decision tree algorithms outperform with 86.6% accuracy for the prediction of future values for 10 Year Dataset and 87.5% for 15 Year Dataset. Meanwhile, it is very uncertain to refrain from the findings of concurrent studies on real-world investing strategies due to the lack of predictable results. The author has concluded that the Decision tree algorithm has the highest prediction accuracy of 89.4% in 20-year dataset compared to other machine learning algorithms and the simple architecture of the Neural Fuzzy Inference System performs with high precision and accuracy with low lagging issues.