Common Pitfalls When Applying Machine Learning

There have been significant advances in machine learning over the last twenty years, which have produced more accurate predictions in a variety of well-validated industrial case studies. On the other hand, there are many ways to incorrectly apply machine learning methods, usually resulting in models that are much less accurate than believed and have poor predictive value. This presentation describes a half dozen common pitfalls that occur when applying machine learning, and ways to avoid such pitfalls. Following these recommendations produce machine learning models that have higher predictive value.