Pandas Feature request: drop all rows containing value

When chaining multiple calls I often want to select certain elements. Unfortunately getting only the values requires assignment and thus breaks the chaining: df_sub = df[df == x]

One hack is: df.replace({x: np.nan}).dropna()

I think a drop_values() function would be nice addition where I could say: df.drop_values(x).more().chaining()

Comment From: jreback

you mean like this?

In [18]: df = pd.DataFrame({'A': range(3), 'B': list('abc')})

In [19]: df.assign(bar=df.A+1).loc[lambda df: df.B=='a']
Out[19]: 
   A  B  bar
0  0  a    1

Comment From: jreback

or dropping

In [20]: df.assign(bar=df.A+1).loc[lambda df: df.B != 'a']
Out[20]: 
   A  B  bar
1  1  b    2
2  2  c    3

Comment From: twiecki

TIL you can supply functions to .loc[]. Yes, exactly like that. Since that works not sure more syntactic sugar is needed although it would simply the process a little bit.

Comment From: jreback

more API is not needed here, this is a pretty common pattern, docs are here: http://pandas.pydata.org/pandas-docs/stable/indexing.html#selection-by-callable

Comment From: twiecki

Thanks @jreback, and sorry for the distraction.