Implicit conversion from float -74.81012658227849 to int loses precision