Dear Jorge,

usually, using the ANN Toolbox from Matlab you can define three datasets: training, validation and test. The training dataset is used to update the weights of the ANN, the validation dataset is used to calculate the error that will control the training process. The best weights are taking for the minimal validation error. Finally, the test dataset is used to evaluate the performance of your ANN.

Another way to identify the accuracy of you predictors is use a dataset defined for yourself. For this, I think you can use the command lines to call your trained ANN, and so calculate an error index, such as MSE, RMSE, MAE, etc. If you are using the graph interface of ANN toolbox, I am not sure if you are able to export/save your trained ANN, but you can do exactly the same by command lines.

I hope my answer can help you. If you have other questions, send me an email.

Kind Regards!

Thanks Bruno.

Hope you are doing good.

Greetings,

Jorge

Dear Jorge Pesantez ,

You can check the testing or validation accuracy of your model. Please consider the issue of over-fitting and bias. You can take help from the following link for performance measurement.

https://en.wikipedia.org/wiki/Confusion_matrix