Accuracy Paradox. How not to fall into the trap of unbalanced dataset

quite interesting article about pitfalls that data scientists may face while working with data