If you've made changes to your data and would like to re-train your AI anomaly detector to take advantage of extra context information, you'll need to erase and rebuild your model.

This step is optional, but recommended when the AI you're rebuilding has several different training data samples.

When signed in as an Administrator, open the anomaly display from the asset dashboard.

Then click Training Data and Download Sample Times.

This downloads a CSV file containing the start and end times of each historical period you're using to train your AI model.

You can erase an AI anomaly detector model by signing in as an Administrator and choosing AI Anomaly Detector in the asset dashboard.

Then press the Erase Model button to clear the AI.

Follow the normal process of creating a new AI.

This step is only needed if you backed up your sample times in step 1.

Next, choose Training Data and then Upload Sample Times.

Choose the file you downloaded in step 1 and press 'Add Training Data'.

This will then upload the list of sample times - the addon will then re-download the training data and re-train the AI.