Should it be possible to interpolate from other features in the imputing. impute_all_assets_by_correlation function?

[achenry]

It looks like whatever reference_col argument is passed to imputing. impute_all_assets_by_correlation, the reference_col then passed to the internally called impute_data function is set to impute_col...

OpenOA/openoa/utils/imputing.py

Line 211 in 47ca939

reference_col=impute_col,

[ejsimley]

Hi @achenry, you're right that reference_col should be passed to the reference_col argument of impute_data instead of impute_col. So that is a quick fix. I also noticed that when determining the most highly-correlated asset to use for imputing, the correlation coefficients between assets are determined using impute_col only:
corr_df = asset_correlation_matrix(data, impute_col)

I think for reference_col to get used as intended, the correlation coefficients should be based on the correlation between reference_col on one asset and impute_col on the other. Does that make sense? I'll try to modify asset_correlation_matrix to give the reference-to-impute column correlation.