MODEL VALIDATION THROUGH THE LINEAR-REGRESSION FIT TO ACTUAL VERSUS PREDICTED VALUES

Validation of predictive models in agriculture is often carried out wi th the simultaneous F test of zero intercept and unit slope when regre ssing predicted outputs on real-system values or vice versa. Such a te st is not suitable for this intended use, because it does not take int o consideration the behaviour of the non-controlled part (the lack of fit) of such models. Moreover, the test is not informative when severa l models are available and consequently, it would nor be able to pick up the best one. A correct proof for model validation should be based on the mean square error of the models being evaluated. (C) 1998 Elsev ier Science Ltd. All rights reserved.