5

I've built an entity set and one of the tables in this entity set is called "inspections". I've set the time_index column for this table, but when running dfs, I'm getting the warning "Using training_window but last_time_index is not set on entity inspections". The documentation shows that this should be set as a series: last_time_index (pd.Series) – Time index of the last event for each instance across all child entities.

Can someone please provide an example of how and what values I should set the last_time_index to?

Note, the calculations are still correct, but I'm concerned about the warning message.

Nick Bernini
  • 121
  • 4

1 Answers1

6

You can add the last_time_index to all entities automatically by calling your_entityset.add_last_time_indexes(). Run this as a last step after your entire entity set is configured.

The last_time_index is only used when you are calculating features with a training window, so we don’t spend the time to create it by default. If you run that command, the warning will go away.

The last_time_index is used so we can exclude rows that have their last data before the start of the window of training data. This is different than time_index, which makes sure the row was known or had its first data before the end of the window of training data.

Max Kanter
  • 2,006
  • 6
  • 16