ValueError: Got an unexpected argument: categories about fma

[drscotthawley]

Fixed... Apparently there's been a change in Pandas, at least for version 0.25.0 (the requirements.txt file for fma doesn't specify a version for pandas).

So the fix is to modify utils.py so that it includes

from pandas.api.types import CategoricalDtype

and the offending line about categories that used to read as

                'category', categories=SUBSETS, ordered=True)

now reads as

                CategoricalDtype(categories=SUBSETS, ordered=True))

. :-)

[andimarafioti]

What does:

import pandas
print(pandas.__version__)

prints?

[andimarafioti]

This error also appears in the creation notebook

[mdeff]

Thanks for reporting! This behavior was indeed removed in pandas 0.25:

Removed the previously deprecated ordered and categories keyword arguments in astype (GH17742)

I used pandas==0.21.0 when developing this code. I will add package versions to the requirements.txt.

[717101]

Hi, I tried to use the new code but the error persist in the next line.
Thanks for your work.

---

TypeError                                 Traceback (most recent call last)
<ipython-input-31-4224f39de5bf> in <module>()
      1 AUDIO_DIR = os.environ.get('fma_small')
----> 2 tracks = utils.load('fma_metadata/tracks.csv')

/content/utils.py in load(filepath)
    212             tracks[column] = tracks[column].astype('category')
    213 
--> 214         return tracks
    215 
    216 

TypeError: astype() got an unexpected keyword argument 'categories'

[717101]

Thanks, I didn't realized that my pandas version was downgraded.

[mdeff]

I tried to use the new code but the error persist in the next line.

The line number of the error (--> 214 return tracks) seems to indicate you're not using the latest version. The latest has a try-catch block for this case:

212        try:
213            tracks['set', 'subset'] = tracks['set', 'subset'].astype(
214                    'category', categories=SUBSETS, ordered=True)
215        except ValueError:
216            # the categories and ordered arguments were removed in pandas 0.25
217            tracks['set', 'subset'] = tracks['set', 'subset'].astype(
218                     pd.CategoricalDtype(categories=SUBSETS, ordered=True))

So both old (pandas<0.25) and new (pandas>=0.25) pandas should work.

Could you try a fresh git clone?

[albert239825]

Hello, I know I'm a little late, but I would like to add that I am still getting the error despite the try-catch block.

Edit: I fixed it by removing the except ValueError: and just changing it to except:

[mdeff]

Thanks for reporting @albert239825. Which exception did you get? If TypeError, could you try #47? Which version of pandas (import pandas; print(pandas.__version__)) are you using?

[albert239825]

@mdeff I didn't get a type error. My pandas version is 1.2.2 and the error I get is "astype() got an unexpected keyword argument 'categories'".
Edit: Also, I undid a commit with the fix and now it's broken again...
Edit: Nevermind, changing
except ValueError: to except: fixed the issue as before

[mdeff]

the error I get is "astype() got an unexpected keyword argument 'categories'"

And that's a TypeError. 😉

Hence I'll merge #47.

[albert239825]

@mdeff Thanks! So sorry for my lack of knowledge.

[mdeff]

No problem. 😉 Thank you for following-up on the issue.