BUG: np.argwhere on pandas series

[rossbar]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• (optional) I have confirmed this bug exists on the master branch of pandas.

---

numpy/numpy#15555 reports an issue with np.argwhere on pandas Series. Reporting here for visibility.

MRE:

>>> import numpy as np
>>> import pandas as pd
>>> s = pd.Series(np.random.randn(5), index=['a', 'b', 'c', 'd', 'e'])
>>> np.argwhere(s < 0)

which, with numpy.__version__ ==1.20.0.dev0+046a736 gives:
pd.version == 0.25.3:

FutureWarning: Series.nonzero() is deprecated and will be removed in a future version.Use Series.to_numpy().nonzero() instead
array([[3]])

pd.version == 1.0.5:

ValueError: Length of passed values is 1, index implies 5.

[simonjayhawkins]

Thanks @rossbar for drawing our attention to this issue.

[ArchangeGabriel]

I’ve run into this while trying to execute a code that used to work back in 2019. The workaround for anyone stumbling upon this is thus to .to_numpy() the series in the np.argwhere() call, e.g. for the above example:

>>> np.argwhere(s.to_numpy() < 0)

[mroeschke]

This looks to work on main now. Could use a test

In [18]: >>> import numpy as np
    ...: >>> import pandas as pd
    ...: >>> s = pd.Series(range(5), index=['a', 'b', 'c', 'd', 'e'])
    ...: >>> np.argwhere(s < 3)
Out[18]: 
array([[0],
       [1],
       [2]])

[mcgeestocks]

I can jump in and add a test for this.