Multi label classification returns mean accuracy of almost 0

Question

I have a dataframe that looks like this (but has in reality > 10000 rows):

data = {\'age\': [54, 21, 7, 18],
        \'sex\': [0, 1, 1, 0],
        \'disea