Approximation error is the difference between an exact value, function, or solution and an approximate representation of it. In machine learning and numerical computation, it can arise when a model, algorithm, or simplified representation cannot fully capture the target quantity.
The discrepancy between an exact value and some approximation to it.