Better machine learning by thinking like a statistician. About model interpretation, paying attention to data, and always staying critical.