PuLP LpVariable of binary type giving value 2

Question

So I wrote few conditions

keys=['a','b','c']
var=LpVariable.dicts('var',(keys,[1,2,3]),cat='Binary')

for key in keys:
   prob+=lpSum([var[key][j] for j in [1,2,3]])==2

Even though I am getting optimal result, for some keys I see that var is taking value 2.0. Why this is happening and is there any solution to it?