Validation is the process of testing the degree to which a model is useful for making forecasts. The sample data are often split into two segments, one being used to estimate the parameters of the model, and the other being used as a holdout set to test the forecasts made by the model. There are many variations on this process of validation.