National level model summary with holdout group: holdout_id indexing error

[j6baptiste]

Hi, I am running into an indexing error while generating an HTML summary for a national model trained with a single knot and an holdout group given as a list of booleans of dimension (n_times,).

The error happens on that line of the source code:

meridian/meridian/analysis/analyzer.py

Line 3764 in fa61ba6

holdout_id = holdout_id[:, time_mask]

When running:

mmm_summarizer.output_model_results_summary(f'meridian_output{model_suffix}.html', filepath, start_date, end_date)

And raises:

IndexError: too many indices for array: array is 1-dimensional, but 2 were indexed

According to the documentation the expected shape for holdout_id in the model specifications for a national level model is (n_times,). But the indexing in the source code is always 2 dimensional.

Could someone provide guidance on how to avoid this issue?

(I can provide traceback / more details about the code if necessary)

[cpulavarthi]

Hello @j6baptiste,

Thank you for bringing this issue to our attention. This bug has been resolved in commit #307fa31 and your code should now work without any errors.

Feel free to reach out if you have any queries regarding Meridian.

Thank you

Google Meridian Support Team