2
$\begingroup$

I am using autoencoder for anomaly detection in warranty data. It is unsupervised. I calculate the reconstruction error by the model and the records with high reconstruction error value is considered as an anomaly. I would like to know, if it is necessary to train/test-split the data.

Any help is much appreciated!

$\endgroup$
1
$\begingroup$

Yes it is still necessary, you are fitting your model on that data and learning it to find a good representation for that sample. Validating whether or not this was actually an anomaly is a lot more difficult then.

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.