jolespin / ensemble_networkx Goto Github PK

I've been getting an error when my X (pd.DataFrame) has an index name.

ValueError: coordinate Samples has dimensions ('16S_ID',), but these are not a subset of the DataArray dimensions ['Samples', 'Edges', 'Statistics']

My X.index.name is causing the error:

X.index.name
# '16S_ID'

The way around this is to set the name to None

X.index.name = None

It seems like there is currently no way to robustly do correlations on binary data (e.g. correlations between two dichotomous categorical variables).
I'll try to implement the Matthews Correlation Coefficient (also known as the Phi Coefficient) in the EnsembleAssociationNetwork class to resolve this.

I initially posted this issue on StackOverflow (https://stackoverflow.com/questions/76715515/how-to-create-xarray-dataset-with-a-pandas-multiindex-for-one-of-the-dimensions). The gist of the issue is that any Pandas MultiIndex objects on Pandas DataFrame or Pandas Series for X or y used in the SampleSpecificPerturbationNetwork().fit method will throw the following error: MissingDimensionsError: cannot set variable 'Samples' with 2-dimensional data without explicit dimension names. Pass a tuple of (dims, data) instead.

You can't convert them to tuples either, the work around for the time being is to convert them to a string until I figure out a better solution.