How to do Multi-Column from_tuples?

Assign direct to columns with the result from pd.MultiIndex.from_tuples passing in your existing columns:

In [186]:l=[('A', 'a'),  ('A', 'b'), ('B','a'),  ('B','b')]df = pd.DataFrame(np.random.randn(5,4), columns = l)dfOut[186]:     (A, a)    (A, b)    (B, a)    (B, b)0 -0.876353  0.553742  1.631858 -0.5613091  0.463058 -0.455014 -0.491336 -1.4360592  0.337810  0.233624 -0.571749 -2.2597633  1.073057 -0.475894  0.999643 -0.3797434  0.441800  0.311202 -0.191552  0.291268In [187]:    df.columns = pd.MultiIndex.from_tuples(df.columns, names=['Caps','Lower'])dfOut[187]:Caps          A                   B          Lower         a         b         a         b0     -0.876353  0.553742  1.631858 -0.5613091      0.463058 -0.455014 -0.491336 -1.4360592      0.337810  0.233624 -0.571749 -2.2597633      1.073057 -0.475894  0.999643 -0.3797434      0.441800  0.311202 -0.191552  0.291268

note that you can assign directly to names attribute of the columns attribute like the following:

df.columns.names = ['Caps','Lower']

not to be confused with the name attribute

Another solution is use MultiIndex.from_tuples with parameter names:

import pandas as pddf = pd.DataFrame({'Value': [1,2,3]}, index=[('A','a'),('B','a'),('B','b')])print (df)        Value(A, a)      1(B, a)      2(B, b)      3df.index = pd.MultiIndex.from_tuples(df.index, names=['Caps','Lower'])print (df)            ValueCaps Lower       A    a          1B    a          2     b          3

This same works with columns, see Edchum's answer:

df.columns= pd.MultiIndex.from_tuples(df.columns, names=['Caps','Lower'])