What is the Dask equivalent of the Pandas .filter() attribute?

Question

I am trying to make sub-DataFrames from a larger DataFrame in Dask. I realize that a lot of the tools found in Pandas used to manipulate DataFrames are present in Dask, however the devs are very transparent about what is not.

One such tool is the df.filter() attribute for DataFrames. I have code that looks like this:

comp2phys = df.filter(FOI[1:], axis=1)

where df is a DataFrame and FOI is a list containing the "fieldnames of interest".

I get this error when I try to filter the dataframe (as I explained above, it is not a tool in Dask)

AttributeError: 'DataFrame' object has no attribute 'filter'

Is there a tool in Dask that will allow me to do this? Thanks!

EDIT

As @ayhan pointed out, this equivalent to df[FOI[1:]]

Isn't this equivalent to `df[FOI[1:]]` (which should work on dask dataframes as well)? — ayhan, May 09 '18 at 19:58
I facepalmed after posting this question because I realized that as well. Thank you for pointing that out Rafael. — Dave L, May 09 '18 at 20:23
(you may want to put that as an answer, so that this question doesn't appear unanswered) — mdurant, May 13 '18 at 01:17
@mdurant are you planning to implement `filter` at some point? — rpanai, Nov 30 '18 at 12:06

score 1 · Answer 1 · answered Nov 30 '18 at 02:36

1

As @ayhan pointed out, this equivalent to df[FOI[1:]]

answered Nov 30 '18 at 02:36

Dave L

77
11

What is the Dask equivalent of the Pandas .filter() attribute?

1 Answers1