Library: | nn |
See also: | rbftrain rbfpredict rbfsave rbfload rbfinfo rbftrain2 |
Quantlet: | rbftest | |
Description: | tests the given rbfnet network |
Usage: | {netOut,AED,MSE} = rbftest(x,y,rbfnet) | |
Input: | ||
x | (n x p) matrix, predictor variables | |
y | (n x q) matrix, response variables | |
rbfnet | composed object (list), RBF network as computed by rbftrain or rbftrain2 | |
Output: | ||
netOut | (n x q) matrix, output of the rbfnet | |
AED | (n x q) matrix, absolute error difference, i.e. abs(netOut - y) | |
MSE | scalar, mean squared error |
library("nn") ; load the library ; training set randomize(2020) n = 10 xt = normal(n,2)+#(-1,-1)' | normal(n,2)+#(+1,+1)' yt =(matrix(n)-1)|matrix(n) ; build the RBF network clusters = 2 learn = 0.1|0.2|0.1 epochs = 5|5 mMSE = 0.05 activ = 0 rbfnet = rbftrain(xt,yt,clusters,learn,epochs,mMSE,activ) ; testing set randomize(1140) x = normal(n,2)+#(-1,-1)' | normal(n,2)+#(+1,+1)' ; predict the output y = rbfpredict(x,rbfnet,0,1) sum((y > 0.5) != yt) ; test the predicted values test = rbftest(x,y,rbfnet) test.MSE
Contents of t [1,] " An 2 - 2 - 1 RBF-network" [2,] " training epochs: 5 - 5" [3,] " cluster's learning rates: 0.2000 - 0.1000" [4,] " output's learning rate: 0.1000" [5,] " minimum mean squared error: 0.050" [6,] " BINARY sigmoid activation function" [7,] " minimum MSE reached: 0.190974 " Contents of sum [1,] 1 Contents of MSE [1,] 0.0030565