Confusion over pred_hdi_upper_94 values

[paul-kinase]

In a synthetic control experiment following this example notebook: https://causalpy.readthedocs.io/en/stable/notebooks/multi_cell_geolift.html

The pred_hdi_upper_94 outputted from the SyntheticControl.get_plot_data() method is simply "treated_agg".

Does this mean that the upper threshold of the credible interval, is simply the difference between my prediction (i.e the synthetic control) and the observed outcome.

Should the prediction - outcome not be near the highest probability density, i.e close to the mean in the posterior distribution of the causal impact, with higher and lower thresholds either side?

@juanitorduz @drbenvincent

🙏

[drbenvincent]

Sorry for the delay @paul-kinase - things have been ultra busy and I've not had much time for open source work.

It took me a while to understand, I thought you meant the values of the pred_hdi_upper_94 column was equal to the _values_ of the treated_agg. So a pedantically more precise description is the values of the pred_hdi_upper_94column are literally the string"treated_agg"`.

Definitely a bug (not my fault). I'll work on a fix for it.

[paul-kinase]

Thank you! @drbenvincent