survey has columns gender and purchased.

Return the same contingency table as the counts version, but with each row scaled to percentages summing to 100, rounded to 2 decimal places.

The normalisation is over rows, not the whole table, so each gender's row sums to 100 regardless of how many people are in it.

Input

  gender purchased
0      f       yes
1      f        no
2      m       yes
3      m       yes
4      m        no
5      f       yes

Output

purchased     no    yes
gender                 
f          33.33  66.67
m          33.33  66.67