df.first_valid_index() fails if there is no valid entry.

[bjlobo]

pandas/pandas/core/frame.py

Lines 3770 to 3777 in 3a7f956

	def first_valid_index(self):
	"""
	Return label for first non-NA/null value
	"""
	if len(self) == 0:
	return None
	return self.index[self.count(1) > 0][0]

import pandas as pd
import numpy as np

df = pd.DataFrame({'a': [np.nan,np.nan]})
df.first_valid_index()

This code results in the following error message, which I believe is called because the boolean series returned by self.count(1) > 0 is all False.

Traceback (most recent call last):
  File "<pyshell#10>", line 1, in <module>
    df.first_valid_index()
  File "C:\Python27\lib\site-packages\pandas\core\frame.py", line 3859, in first_valid_index
    return self.index[self.count(1) > 0][0]
  File "C:\Python27\lib\site-packages\pandas\indexes\base.py", line 1423, in __getitem__
    return getitem(key)
IndexError: index 0 is out of bounds for axis 0 with size 0

Should there be a check where None is returned if self.count(1) > 0 is all False?
As an aside, this issue also occurs in df.last_valid_index() for the same reason.

[gfyoung]

@bjlobo : Thanks for reporting this!

I think returning None is fine. Make sure to update the docs + add test when you do so.

[jreback]

yes this should act similarly to Series in this condition.

[bjlobo]

@gfyoung I have never modified the code or documentation or written tests before and I am not sure how to start doing this.

[cggarvey]

Sorry, you can disregard. I just noticed skwbc's PR above. Should've had coffee before looking at the tracker...