We'll cut down in the size of the DataFrame we're working with by throwing out weapon data for weapons with indices over three (which are really only present on ships anyway).