convert conditional dtype to corresponding integer of floating point assignment & fix test