In an ideal world of data, we would always have a classification problem where the data classes are even or balanced, that is each class will have an equal count of instances. Unfortunately, there are several classification problems where the data can be very highly imbalanced.