Does new input data (test data) need to be 'manually' standardized before passing it to a trained Regression Neural Network model?

Hi all,

I trained a neural network model using the Regression Learner App in MATLAB, and I set "Standardize" to true. According to the documentation, that option standardizes the predictors so that they have mean 0 and standard deviation 1.

Now I want to test this model with new data which has never been seen by the model before, and I wonder if I need to apply the standardization to this data 'manually' (by using the N = normalize(A) function, for example) before passing it to the trained model to predict the ouputs, or if this is automatically performed by the trained model everytime it receives any input.

I couldn't find anything about it in the documentation so any help on the topic would be greatly appreciated :)

Answer 1

[Anshika Chaurasia]

Hi,

There is no need to standardize test data before passing it to trained Regression Neural Network.

Moreover, in the documentation, it is mentioned that Regression Learner uses the fitrnet function to train the neural network models.

In the examples mentioned in fitrnet documentation, you can see to evaluate the performance of regression model on the test set, no normalization (or standardization) is done before.