This message was deleted.
# hamilton-helps
Slackbot
01/14/2023, 5:32 PMThis message was deleted.
e
Elijah Ben Izzy
01/14/2023, 5:44 PMSo the spine_column is done at the end to join the series. Are you just using a single dataframe? Or joining multiple in the results builder?
a
Anwar Brini
01/14/2023, 5:46 PMi m joining multiple dataframes
Anwar Brini
01/14/2023, 5:46 PMBut my output is a single dataframe
e
Elijah Ben Izzy
01/14/2023, 5:47 PMGot it
Elijah Ben Izzy
01/14/2023, 5:47 PMSo you’re joining as part of your transforms, not the results builder.
Elijah Ben Izzy
01/14/2023, 5:47 PMI think you can pass any valid series name from the final result although I’d have to dig in to be 100% sure
a
Anwar Brini
01/14/2023, 5:47 PMyes
Anwar Brini
01/14/2023, 5:47 PMokay i see
e
Elijah Ben Izzy
01/14/2023, 5:47 PME.G. the join is going to be a no-op for you
Elijah Ben Izzy
01/14/2023, 5:48 PMit might be that you want to return that column as well — I can look after I fix the typing piece 🙂 Testing as we speak…
a
Anwar Brini
01/14/2023, 5:49 PMokay cool thank youu !!
e
Elijah Ben Izzy
01/14/2023, 5:50 PMYeah! So took a quick look — at the end we just join it based on the output — its pretty simple:
Copy code
def build_result(
self, **outputs: typing.Dict[str, typing.Any]
) -> typing.Union[pd.DataFrame, ps.DataFrame]:
# we don't use the actual function for building right now, we use this hacky equivalent
df = ps.DataFrame(outputs[self.spine_column])
for k, v in outputs.items():
<http://logger.info|logger.info>(f"Got column {k}, with type [{type(v)}].")
df[k] = v
if isinstance(self.result_builder, base.PandasDataFrameResult):
return df.to_pandas()
else:
return dfElijah Ben Izzy
01/14/2023, 5:50 PMIts possible that it’ll break with a dataframe result — I’ll need to test it ou. D.G. df[k] = v appears to be expecting a series.
s
Stefan Krawczyk
01/14/2023, 5:50 PM@Anwar Brini yeah it’s possible this way of building a result isn’t what you want.
e
Elijah Ben Izzy
01/14/2023, 5:53 PMYeah — @Stefan Krawczyk has a point. You have no need for the joins if you’re operating on DFs as opposed to columns…
s
Stefan Krawczyk
01/14/2023, 5:53 PMif you let us know logically what you want to happen — we can help adjust/augment the code.
Stefan Krawczyk
01/14/2023, 5:53 PMat the very least give you a custom adapter that does what you want.
a
Anwar Brini
01/14/2023, 5:56 PMI’m currently doing join/transformation operations in the module and some transformations to the final dataframe that i want as an output. I ve tried using the index column of this dataframe as a spine column but it didin’t work and even with extract_columns in the last function that didin’t seem to work either
s
Stefan Krawczyk
01/14/2023, 5:58 PMdo you have time to jump on a call now? might be easier to see some code/chat live?
a
Anwar Brini
01/14/2023, 5:58 PMyeah ofc
s
Stefan Krawczyk
01/14/2023, 7:32 PM@Anwar Brini the branch now has the fix for the parameter definition. When you get a chance could you validate please? 🙂