/Fuzzy_Systems_Regression

TSK models that are using the hybrid method for regression problems

Primary LanguageMATLABMIT LicenseMIT

Fuzzy_Systems_Regression

TSK models that are using the hybrid method for training. For the task 1 use Airfoil Self-Noise Data Set. For the task 2 use Superconductivty Data Set

Task 1 Model 1

  • Use grid partition
  • Change the output function to constant
  • Use 2 membership function
  • Use gbellmf as membership function type
  • Train the TSK model with hybrid method
  • Evaluate the model
    • R2
    • RMSE
    • NMSE
    • NDEI

Task 1 Model 1

  • Use grid partition
  • Change the output function to constant
  • Use 3 membership function
  • Use gbellmf as membership function type
  • Train the TSK model with hybrid method
  • Evaluate the model
    • R2
    • RMSE
    • NMSE
    • NDEI

Task 1 Model 3

  • Use grid partition
  • Change the output function to linear
  • Use 2 membership function
  • Use gbellmf as membership function type
  • Train the TSK model with hybrid method
  • Evaluate the model
    • R2
    • RMSE
    • NMSE
    • NDEI

Task 1 Model 4

  • Use grid partition
  • Change the output function to linear
  • Use 3 membership function
  • Use gbellmf as membership function type
  • Train the TSK model with hybrid method
  • Evaluate the model
    • R2
    • RMSE
    • NMSE
    • NDEI

Task 2

  • Use Subtractive Clustering
  • Create a grid search for the best number of features and values of radii
  • Use relieff for feature selection
  • Compare metrics between models to find the best parameters