K-FOLD CROSS-VALIDATION (BATCH)

K-FOLD CROSS-VALIDATION TO TRAIN RADIAL-BASIS NEURAL NETWORK MODEL IN MATLAB

Program to find Network parameter of a Radial-basis neural network model MATLAB function 'newrb' is used to train a neural network for a given data set.
The function has form net = newrb(Y, X, goal,spread).
spread (a network parameter) will affect the model performance.

The program can be used to obtain best value of 'Spread constant' for a given data set. A box-plot showing the variation in cross-validation error with spread constant is obtained. A designer may then analyse the plot to figure out the optimum 'spread constant'.