pandas apply: Expression of type "Series[Unknown]" cannot be assigned to return type "DataFrame"

[tekumara]

Environment data

• Language Server version: v2022.7.20 (pyright 7cadfb74)
• OS and version: macOS
• Python version (& distribution if applicable, e.g. Anaconda): python 3.9, pandas 1.4.3

Code Snippet

def normalise(df: DataFrame) -> DataFrame:
    return df.apply(lambda x: x)

Repro Steps

Expected behavior

No type error

Actual behavior

Expression of type "Series[Unknown]" cannot be assigned to return type "DataFrame"
  "Series[Unknown]" is incompatible with "DataFrame" Pylance[reportGeneralTypeIssues]

Logs

XXX

[bschnurr]

there are two overloads for that apply function

could convert to dataframe

def normalise(df: DataFrame) -> DataFrame:
    return df.apply(lambda x: x).to_frame()

or call the other overload that returns a Dataframe

def normalise(df: DataFrame) -> DataFrame:
    return df.apply(lambda x: x, 0 )

[tekumara]

Oh interesting, thanks @bschnurr both of those work, and I've learnt more about the pandas type stubs which is great!

However it is non-idiomatic usage of df.apply (at least in our team). Do you know if the type stubs can be improved to address this, or is it impossible to express?

[bschnurr]

stubs are pulled from here https://github.com/pandas-dev/pandas-stubs

checkout their issues related to apply

and try asking them about usage.

[bschnurr]

closing since this is related to pandas library itself

[bschnurr]

better fix. see https://github.com/microsoft/pylance-release/issues/3491#issuecomment-1282691441