You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
#363 adds pandas as a dependency and in the tutorials we convert dicts of dicts to dataframes anyway, therefore we could simply return some properties as dataframes. @amueller@janvanrijn what's your opinion on this?
Nope, this is for function like list_dataset which could naturally be represented as dataframes. Check this example to see which lines could be saved by returning dataframes by default.
I'm also in favor of (adding the option of) returning a dataframe. I am probably in favor of making it the new default behavior. With pandas becoming the main format of representing (data of the) datasets, it makes sense to also represent the different views with a dataframe.
Activity
PGijsbers commentedon Mar 18, 2019
Should we close this issue as this is possible because of #548, or after the release of 0.9 when dataframes are returned by default?
mfeurer commentedon Mar 18, 2019
Nope, this is for function like
list_datasetwhich could naturally be represented as dataframes. Check this example to see which lines could be saved by returning dataframes by default.However, I should not decide this by myself. @janvanrijn @joaquinvanschoren @amueller @PGijsbers @ArlindKadra any comments/agreements/disagreements?
joaquinvanschoren commentedon Mar 18, 2019
I would like to have the option to get a dataframe directly. It makes some code a lot shorter and cleaner.
PGijsbers commentedon Mar 18, 2019
I'm also in favor of (adding the option of) returning a dataframe. I am probably in favor of making it the new default behavior. With pandas becoming the main format of representing (data of the) datasets, it makes sense to also represent the different views with a dataframe.
janvanrijn commentedon Mar 19, 2019
Agreed.