How to make a pandas crosstab with percentages?

From Pandas 0.18.1 onwards, there's a normalize option:

In [1]: pd.crosstab(df.A,df.B, normalize='index')
Out[1]:

B              A           B           C
A           
one     0.333333    0.333333    0.333333
three   0.333333    0.333333    0.333333
two     0.333333    0.333333    0.333333

Where you can normalise across either all, index (rows), or columns.

More details are available in the documentation.

We can show it as percentages by multiplying by 100:

pd.crosstab(df.A,df.B, normalize='index')\
    .round(4)*100

B          A      B      C
A                         
one    33.33  33.33  33.33
three  33.33  33.33  33.33
two    33.33  33.33  33.33

Where I've rounded for convenience.

pd.crosstab(df.A, df.B).apply(lambda r: r/r.sum(), axis=1)

Basically you just have the function that does row/row.sum(), and you use apply with axis=1 to apply it by row.

(If doing this in Python 2, you should use from __future__ import division to make sure division always returns a float.)