DataFrame.melt fails if there are duplicate value_vars

[swenlaur]

DataFrame.columns.get_indexer fails if there are duplicate column names.

pandas/pandas/core/reshape/melt.py

Line 109 in 05552d3

idx = frame.columns.get_indexer(id_vars + value_vars)

This is really unfortunate -- melting should handle such cases

id value value
id1 2 3

should go to

id1 2
id1 3

[simonjayhawkins]

Thanks @swenlaur for the report.

Please read this guide detailing how to provide the necessary information for us to reproduce your bug.

and can you use the bug template that is available when you open an issue from the GitHub web interface.

[swenlaur]

Minimal code to trigger the bug (regression from earlier versions)

from pandas import DataFrame
DataFrame([['id', 2, 3]]).set_axis(['id_var', 'value_var', 'value_var'], axis=1).melt(id_vars=['id_var'], value_vars=['value_var'])

The resulting data frame contains columns with the same name. This is not illegal as sometimes uncleaned data in this form and the most sane way to clean it is using melt.

[simonjayhawkins]

regression from earlier versions

will add the regression tag for now pending investigation. posting more info on versions could help

[swenlaur]

I do not know the highest package number but repro works with pandas=1.0.0. I guess it stoped working after the multi-level colums patch was added to the code.