outliers() is missing outlier filename column

[dbickson]

When I tried to list the outliers, some error pops up ("AttributeError: 'DataFrame' object has no attribute 'img_filename_outlier'"). I attached the screenshot in the attachment.

I am using version 0.917. It is clear that there is no "img_filename_outlier" in the outlier_df dataframe... (outlier_df = fd.outliers()) So I think that might be a compatibility problem?

Only "outlier" "nearest" "distance" "index" "filename_nearest" "error_code_nearest" "is_valid_nearest" are found.

Thanks

Best,

Zilun

[dnth]

I can confirm having this issue on fastdup version '0.921'

[dbickson]

The solution is to change img_filename_outlier to filename_outlier. @dnth can you please update the docs accordingly?

[dnth]

@dbickson with the current version (0.921) fd.outliers() does not produce a DataFrame with the name filename_outlier nor img_filename_outlier.

Changing img_filename_outlier to filename_outlier would also result in the same error.

[dnth]

In version 0.906, fd.outliers() produces a DataFrame with the img_filename_outlier column

[dbickson]

Thanks @dnth for clarifying! We will fix this

[dbickson]

Fixed in 0.922