Implicit conversion from float -94.46808510638297 to int loses precision