Implicit conversion from float -47.95620437956205 to int loses precision