Detecting and Handling Fake Friend Requests via Machine Learning

meta How to identify, validate, and handle fake friend requests on a social platform? Please elaborate from the perspectives of feature engineering and model training. Provide a solution including how to collect data, model selection, and evaluation metrics. Example Input 5 2 1 request_sent 2 mutual_friend_count 3 friend_acceptance_rate 5 frequent_inactive_response 7 request_count