Query on Series

[max-sixty]

I often find myself building a chain of logic, and then wanting to slice over some expression. Here's a contrived example:

In [12]: pd.Series(range(10)).mul(5).pipe(lambda x: x**2).pipe(lambda x: x-500)
Out[12]: 
0    -500
1    -475
2    -400
3    -275
4    -100
5     125
6     400
7     725
8    1100
9    1525
dtype: int64

...and then having to assign to a separate variable and slice using that. Which is not the worst affliction that could strike western civilization, but can disrupt the flow in a repl a little.

In [13]: s=pd.Series(range(10)).mul(5).pipe(lambda x: x**2).pipe(lambda x: x-500)

In [14]: s[s>200]
Out[14]: 
6     400
7     725
8    1100
9    1525
dtype: int64

Instead, would it be reasonable to extend query or filter to do this. For example:

pd.Series(range(10)).mul(5).pipe(lambda x: x**2).pipe(lambda x: x-500).query('>200')
or
pd.Series(range(10)).mul(5).pipe(lambda x: x**2).pipe(lambda x: x-500).filter(lambda x: x>200)

[jreback]

anything different from #11485 ?

[max-sixty]

No that covers it!

Closing.

FWIW Series.filter seems a bit orphaned - not much in the docs, and its first arg items looks like nothing more than a slice. I wonder if this functionality could come from #11485.

[jreback]

@MaximilianR

yeah .filter is only on labels. its a bit of an odd spelling, as typically .filter is on data. though we use .query.

I think we have an issue about this somewhere.