gh:552 - Corrected Series.mean return type

[ramvikrams]

• Closes Series.mean() does not return an float #552
• Tests added: Please use assert_type() to assert the type of any return value

[ramvikrams]

Now MyPy will fail because of tests in files test_windowing,test_resampler,test_frame fail. So I think either we can suppress the warning or changing the DataFrame.mean() method

[Dr-Irv]

• For the errors in test_windowing.py, change the return types of _mean() to np.float64
• For the errors in test_resampler.py, change the return types of the functions h() in line 161 and f() in line 251 to np.float.64
• For the errors in test_frame.py, change the return types of these functions to np.float64:
  • sum_mean() in line 2046
  • lfunc should have type Callable[[pd.DataFrame], np.float64] in line 2050

[Dr-Irv]

don't change this test. please revert

[Dr-Irv]

above two tests should be using s.mean() and checking result against np.float64

[ramvikrams]

• For the errors in test_windowing.py, change the return types of _mean() to np.float64
• For the errors in test_resampler.py, change the return types of the functions h() in line 161 and f() in line 251 to np.float.64

• For the errors in test_frame.py, change the return types of these functions to np.float64:

  • sum_mean() in line 2046
  • lfunc should have type Callable[[pd.DataFrame], np.float64] in line 2050

• for test_resampler.py
  Yes, I had done that on my local system aready but the problem is after doing this change, in h() has incompatible types and then same in f()
• for test_frame
  Same for both of these 2 also they had incompatible types

[Dr-Irv]

• For the errors in test_windowing.py, change the return types of _mean() to np.float64
• For the errors in test_resampler.py, change the return types of the functions h() in line 161 and f() in line 251 to np.float.64

• For the errors in test_frame.py, change the return types of these functions to np.float64:

  • sum_mean() in line 2046
  • lfunc should have type Callable[[pd.DataFrame], np.float64] in line 2050

• for test_resampler.py
  Yes, I had done that on my local system aready but the problem is after doing this change, in h() has incompatible types and then same in f()
• for test_frame
  Same for both of these 2 also they had incompatible types

You may have to make changes to other parts of the tests to make this work, since the tests assumed float was the result of Series.mean(), and now you are changing it to np.float64. Feel free to do that, and I will review.

[ramvikrams]

You may have to make changes to other parts of the tests to make this work, since the tests assumed float was the result of Series.mean(), and now you are changing it to np.float64. Feel free to do that, and I will review.

Thanks

[ramvikrams]

Sir I actually what should we write the assert_type when it is unknown. Means the assert_type is Unknown itself.

[Dr-Irv]

Sir I actually what should we write the assert_type when it is unknown. Means the assert_type is Unknown itself.

If the results are coming back as Unknown from the type checker, it means there is a bug somewhere in the stubs that you'll have to figure out.

[ramvikrams]

If the results are coming back as Unknown from the type checker, it means there is a bug somewhere in the stubs that you'll have to figure out.

thanks i'll do it

[ramvikrams]

The test_resampler errors are now gone but could you suggest a fix for test_frame I could not get it

[Dr-Irv]

The test_resampler errors are now gone but could you suggest a fix for test_frame I could not get it

Can you try the following?

• Add np.integer | np.float to the type IndexIterScalar in _typing.pyi .

If that creates new typing errors, then change the callables for groupby.apply() that declare Scalar as a return type to also allow np.integer | np.float as a return type.

[ramvikrams]

Done Sir

[Dr-Irv]

thanks @ramvikrams