My question is nearly identical to

Finding frequency of each value in all categorical columns across a dataframe

but I need the probabilities, instead of the frequencies. We can use the same example dataframe:

df = pd.DataFrame(
    {'sub_code' : ['CSE01', 'CSE01', 'CSE01', 
                   'CSE02', 'CSE03', 'CSE04',
                   'CSE05', 'CSE06'],
     'stud_level' : [101, 101, 101, 101, 
                  101, 101, 101, 101],
     'grade' : ['STA','STA','PSA','STA','STA','SSA','PSA','QSA']})

I tried adapting this answer

/a/70811258

in the following way:

out = (df.select_dtypes(object)
       .melt(var_name="Variable", value_name="Class")
       .value_counts(dropna=False, normalize=True)
       .reset_index(name="Probability")
       .sort_values(by=['Variable','Class'], ascending=[True,True])
       .reset_index(drop=True))

However, the code doesn’t work, because the sum of the class probabilities for each variable is not 1. What am I doing wrong?