Pandas sudden error in code

Question

I have this df:

A B C D a d a z b c z c

My goal is to pull out the column names where df == 'z' so I used:

df['ColumnNames'] = df.where(df=='z' ).stack().reset_index().groupby('level_0')['level_1'].apply('+ '.join)

It used to work fine; however today I got an error saying: "TypeError: Could not compare ['z'] with block values"

Does anyone know why this happens? Many thanks!

I wasn't able to reproduce the error you have. Did you make any changes to your python packages in recent days? Have you tried to re-run your code or restart your jupyter notebook? What's your python and pandas version? — Yilun Zhang
– Yilun Zhang, Commented Feb 27, 2018 at 21:19
My guess is that you have a version of pandas that is borking when comparing int to str. Try df.astype(object) == 'z' — piRSquared
– piRSquared, Commented Feb 27, 2018 at 21:20
Thank you for all the help. I pulled the data from SQL maybe something changed in the back end but I was about to fix it now. — TylerNG
– TylerNG, Commented Feb 27, 2018 at 21:25
@piRSquared feel free to modify it :-) make it community :-) — BENY
– BENY, Commented Feb 27, 2018 at 21:30

2 revs · Accepted Answer · 2018-02-27 21:29:50Z

Your code working fine on my side, check whether contain space in your df , also we short it with

df[df=='z'].stack() Out[216]: 0 D z 1 C z dtype: object df[df=='z'].stack().index.get_level_values(1) Out[218]: Index(['D', 'C'], dtype='object')

From the Comment above by PiR

My guess is that you have a version of pandas that is borking when comparing int to str. Try df.astype(object) == 'z'

it wasn't a space or anything. I think it must be one of the data type problem. I was about to fix it now with df.astype(str) =='z'. It's weird because I just got the problem today but before it was fine.
@TylerNG aha , That is ok , I will make this answer to community and adding Pir 's solution

1 Answer 1