Premium
When is it safe to extend a prediction equation?—An answer based upon factor and discriminant function analysis
Author(s) -
Wallis James R.
Publication year - 1967
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/wr003i002p00375
Subject(s) - linear discriminant analysis , discriminant function analysis , discriminant , principal component analysis , population , function (biology) , statistics , optimal discriminant analysis , mathematics , econometrics , computer science , artificial intelligence , demography , evolutionary biology , sociology , biology
Prediction equations for hydrologic events developed from one population of observations (watersheds) are often solved for another population that is removed either in time or in space. Predictions of this kind are never certainties, although some predictions are obviously more uncertain than others. This paper proposes an empirical uncertainty classification that may be found useful for separating probably successful from probably unsuccessful extensions of prediction equations. The classification system is illustrated by a prediction equation for suspended sediment discharge developed from some watersheds in California, and by a discriminant function for marine versus nonmarine sediments based upon microelements. (Key words: Statistics; principal components analysis; linear discriminant functions).