Why did you calculate precision and recall instead of giving a prediction whether it is an anomaly or not?

[xuzhang5788]

Thank you for your clear code.

You used X_clean to train AE, and then if the input is X_bad_filtered, your model should classify them as anomalies. I understood your code like this. However, I don't understand why you calculated precision, recall, and F1. If there is a new dataset, how to classify each data as an anomaly or not?

[Rachnog]

Hi @xuzhang5788 , in my case, I had labeled control data for anomalies, so I could calculate precision and recall :)