postreg (Neural Network Toolbox)

Postprocess the trained network response with a linear regression

[m,b,r] = postreg(A,T)

postreg postprocesses the network training set by performing a linear regression between each element of the network response and the corresponding target.

postreg(A,T) takes these inputs,

A - 1 x Q array of network outputs. One element of the network output.

T - 1 x Q array of targets. One element of the target vector.

and returns,

M - Slope of the linear regression.

B - Y intercept of the linear regression.

R - Regression R-value. R=1 means perfect correlation.

Examples

In this example we normalize a set of training data with prestd, perform a principal component transformation on the normalized data, create and train a network using the pca data, simulate the network, unnormalize the output of the network using poststd, and perform a linear regression between the network outputs (unnormalized) and the targets to check the quality of the network training.

p = [-0.92 0.73 -0.47 0.74 0.29; -0.08 0.86 -0.67 -0.52 0.93];
t = [-0.08 3.4 -0.82 0.69 3.1];
[pn,meanp,stdp,tn,meant,stdt] = prestd(p,t);
[ptrans,transMat] = prepca(pn,0.02);
net = newff(minmax(ptrans),[5 1],{'tansig''purelin'},'trainlm');
net = train(net,ptrans,tn);
an = sim(net,ptrans);
a = poststd(an,meant,stdt);
[m,b,r] = postreg(a,t);

Algorithm

Performs a linear regression between the network response and the target, and then computes the correlation coefficient (R-value) between the network response and the target.