Estimating the Error in Model Predictions
Statistical procedures are considered for the problem of determining how well a model performs in predicting the values of variables that are observed in a real system of interest. A careful development of the assumptions underlying a procedure proposed by Freese is presented, along with some extensions and alternate interpretations of this procedure. The uses and interpretations of confidence intervals, prediction intervals, and tolerance intervals for the error involved in using a model are discussed. Two examples using stochastic stand simulation models for predicting stand basal areas are also given. Forest Sci. 30:454-469.
No Supplementary Data
No Article Media
Document Type: Journal Article
Affiliations: Associate Professor of Forestry and Statistics, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061
Publication date: 01 June 1984