Implicit conversion from float -7.214673913043484 to int loses precision