Machine learning models are not perfect; they are prone to a number of errors. The two most common sources of errors are bias and variance. Although two distinct problems, they are interconnected and relate to a model's available degree of freedom or complexity.