Pandas proposal: new sorting order for columns

when I want to sort one of the columns with a specific order, there is the only way to do this right now. example, we have this dataframe

import pandas as pd
df = pd.DataFrame({'good': [10], 'excellent': [20], 'bad': [10]})
df.head()
   bad  excellent  good
0  10   20         10  

and If I want to sort the columns with the good excellent bad order, there is the only way

pd.DataFrame(df, columns=['good', 'excellent', 'bad'])
   good  excellent  bad
0  10    20         10 

but this is ugly.

can you consider this way of solution?

import pandas as pd
df = pd.DataFrame({'good': [10], 'excellent': [20], 'bad': [10]})
df.sort_columns(specific_order=['good', excellent', 'bad'])
   good  excellent  bad
0  10    20         10 

Comment From: sinhrks

I think the usecase is limited, and the following is simpler:

df[['good', 'excellent', 'bad']]

Or, using orderd Categorical as columns meets your requirement. - http://pandas.pydata.org/pandas-docs/stable/categorical.html#sorting-and-order

Finally, an generic idea is adding new arg like standard sort's key arg, not sure how often it'll be used.

Comment From: jreback

@SamuraiT you can do this:

In [3]: df.reindex(columns=['excellent', 'good', 'bad'])
Out[3]: 
   excellent  good  bad
0         20    10   10

or use a Categorical as @sinhrks suggest.

Comment From: SamuraiT

reindex is a good solution, thanks :D