1

I'm trying to perform a multivariate time series anomaly detection. I have training data that consists of "normal" data. I train on this data and detect anomalies on the test set that contains normal + anomalous data. My understanding is that it would be wrong to tweak the model hyperparameters based on the results from the test set.

Question: What would the train/validate/test set look like to train and evaluate a time-series anomaly detector?

siaabd001
  • 11
  • 1
  • Same as for time-series in general: https://stats.stackexchange.com/questions/350655/leave-one-out-cross-validation-for-lstm/351765#351765 – Tim Oct 01 '21 at 22:10
  • Since my goal is anomaly detection, using the typical split may not guarantee that my validation set has anomalies. How can I ensure there are anomalies in my validation set? If I don't need to, then how can I tune the hyperparameters? – siaabd001 Oct 01 '21 at 22:43

0 Answers0