Drop If syntax

[jseabold]

Maybe there's already a way to do this, and there's certainly a safer and more robust/full-featured way to do this, but I just monkey-patched my DataFrames to have a dropif method because I got tired of typing.

def dropif(self, col, cond):
    return self.ix[~eval("self['%s']"%col +cond)]

In the spirit of Stata's drop and keep commands.

[lodagro]

df.drop(df.index[boolean_mask]), is what i use.

In [82]: df = pd.DataFrame(np.random.randn(10,5), columns=list('ABCDE'))

In [83]: df
Out[83]: 
          A         B         C         D         E
0 -1.069739 -1.736850  1.883675 -1.002858  1.055439
1 -0.847116 -0.151453  0.973839  1.183713 -0.466413
2 -1.068765  1.070346  0.109815  0.327656  0.749683
3  1.302732  0.641277  0.212475  1.194309 -0.224311
4  1.411065  0.230903  1.574795  1.823085 -0.244319
5  1.020547  0.233539 -1.386656 -0.511132  0.436428
6 -0.855869 -0.130812 -1.397438 -0.785379  0.823477
7  0.178139  2.027483 -0.810593 -1.196736  0.192493
8  0.341994  0.844972 -0.245320 -2.516069 -1.703321
9  0.698949  0.290125 -0.693364 -2.742538  1.007149

In [84]: df.drop(df.index[(df.A > 0) & (df.E < 0)])
Out[84]: 
          A         B         C         D         E
0 -1.069739 -1.736850  1.883675 -1.002858  1.055439
1 -0.847116 -0.151453  0.973839  1.183713 -0.466413
2 -1.068765  1.070346  0.109815  0.327656  0.749683
5  1.020547  0.233539 -1.386656 -0.511132  0.436428
6 -0.855869 -0.130812 -1.397438 -0.785379  0.823477
7  0.178139  2.027483 -0.810593 -1.196736  0.192493
9  0.698949  0.290125 -0.693364 -2.742538  1.007149

[paulproteus]

Hi @lodagro and @jseabold ,

Is the right conclusion that pandas should close this ticket?

Or should, instead, we keep the ticket open, and suggest that a contributor write the small bit of code it would take to support @jseabold 's request?

It seems somewhat difficult to support this use case cleanly, so best to close the ticket in my opinion. But I thought I'd ping the ticket and see what you all think.

If we believe the issue can be fixed by adding a note in the documentation, that could be a good task for contributors.

[wesm]

Let's leave it open. Fruitful fodder for future design discussions (e.g. more eval? as in "more cowbell")

[petehuang]

A dropif functionality was presented in 0.13.0 via DataFrame.query. Should we close this issue?

Also tagging in @jreback for a more currently active maintainer:

[bashtage]

Is this still relevant?

[jreback]

pretty idiomatic

In [15]: df.loc[~((df.A > 0) & (df.E < 0))]