The performance evaluating criteria of neural network

Hello every one

I used neural network to predict two value

i use this code:

x=xlsread('Features');
t=xlsread('Target');
trainFcn = 'trainbr';  
hiddenLayerSize = 10;
net = fitnet(hiddenLayerSize,trainFcn);
net.divideParam.trainRatio = 70/100;
net.divideParam.valRatio = 15/100;
net.divideParam.testRatio = 15/100;
[net,tr] = train(net,x,t);
y = net(x);
e = gsubtract(t,y); 
performance = perform(net,t,y);

how can i use the performance evaluating criteria (the mean absolute error (MAE), mean square error (MSE), R2 (coefficient of determination), rootmeansquareerror(RMSE),meanabsoluteper- centage error (MAPE), and index ofagreement (IA) values) and how can i know the best network

Answer 1

[Meet]

Hi isra,

The key points for evaluating your neural network model’s performance are:

1. Lower values for "MAE", "MSE", "RMSE", and "MAPE", and higher values for "R²" and "IA" means better performance.
2. Ensure that the network performs well not only on training data but also on validation and test data to avoid overfitting.

You can refer to the resource below for more information:

https://medium.com/analytics-vidhya/mae-mse-rmse-coefficient-of-determination-adjusted-r-squared-which-metric-is-better-cd0326a5697e