Maintaining the order of the categorical variable that is passed into pd.crosstab

[nsriram13]

xref #8731, soln might be the same

Currently when we do a crosstab, the distinct values in each column is reported in the lexical order. But crosstabs are usually useful when we have categorical data (that may have an inherent ordering).

import pandas as pd

d = {'MAKE' : pd.Series(['Honda', 'Acura', 'Tesla', 'Honda', 'Honda', 'Acura']),
'MODEL' : pd.Series(['Sedan', 'Sedan', 'Electric', 'Pickup', 'Sedan', 'Sedan'])}
data = pd.DataFrame(d)
pd.crosstab(data['MAKE'],data['MODEL'])

data['MODEL'] = data['MODEL'].astype('category')
data['MODEL'] = data['MODEL'].cat.set_categories(['Sedan','Electric','Pickup'])
pd.crosstab(data['MAKE'],data['MODEL'])

Both the cross-tab statements above result in the same output as below - essentially the code I believe is performing a lexical sort on the contents of the Series being passed.

Output:
MODEL  Electric  Pickup  Sedan
MAKE                          
Acura         0       0      2
Honda         0       1      2
Tesla         1       0      0

Would it be possible for crosstab to maintain the ordering of the categorical variable if column.cat.ordered on the passed column is True? Thanks!

[jreback]

this sounds sensible. Would you be interested in doing a pull-request for this? should be straightforward

[nsriram13]

I have never worked with pandas code base and have no idea where to look. If you can point me to any resource that can help me get going I can definitely dabble around. But if it is an easy update to make for someone who is more familiar with the code - I would definitely recommend you assign it to them.

[jreback]

https://github.com/pydata/pandas/wiki/Contributing

always a good time to become familar!

lmk know how it goes