Keep original column order

[arjobsen]

Missing functionality
The report sorts the columns alphabetically. Most of the table I use pandas profiling on (for exploratory analysis) have a certain column order with the most important columns first and relevant columns adjacent.

Proposed feature
In the report I'd like to have the original column order retained. This would make inspection of the report much easier and more intuitive.

An option such as keepColumnOrder=true would be nice.

Alternatives considered

Additional context

[sbrugman]

Hi @arjobsen, this is already an option!

df.profile_report(sort="None")

All options can be found in the configuration file (although more explicit documentation would be better):
https://github.com/pandas-profiling/pandas-profiling/blob/master/src/pandas_profiling/config_default.yaml

[arjobsen]

Great, thanks! I searched for it in "the regular documentation" and on Google but couldn't find it.

[mcombalia]

I think the sorting configuration is not working for sort: None, it does work for ascening and descending.

profile = ProfileReport(df, title='Pandas Profiling Report', config_file='config_custom.yaml')
in config_custom.yaml...
sort: None
And the variables are getting sorted in ascending order.

[arjobsen]

None should be a string "None" I think

[sbrugman]

None should be a string "None" I think

This is true when using the sort="None" argument, but is not required in the configuration file.

[mcombalia]

I think the sorting configuration is not working for sort: None, it does work for ascening and descending.

profile = ProfileReport(df, title='Pandas Profiling Report', config_file='config_custom.yaml')
in config_custom.yaml...
sort: None
And the variables are getting sorted in ascending order.

The issue has to do with multiprocessing returning random order of columns, it is already getting looked at in:

#377 (comment)

Thanks