Implicit conversion from float -47.79803646563815 to int loses precision