Application of regression analysis to inventory data with measurements on successive occasions
1993
West, P.W.
Regression analysis is often used to build yield prediction models with inventory data which contain successive measurements from individual forest plots. When ordinary least-squares (OLS) regression is used with such data sets, the parameter estimator is unbiased, but the estimator of the covariance matrix is negatively biased. This prevents the use of normal hypothesis tests of the fitted regression. This paper describes how these problems arise and various approaches that may be used to deal with them. The use of a two-stage model building procedure is advocated to aid in finding an appropriate model form to be used for the regression. It is concluded that the use of permanent sample plots with many remeasurements should be limited in inventory and used primarily to assist in determining an appropriate form for the growth model. Maximising single plot measurements will improve sampling intensity and allow more efficient use of inventory data
显示更多 [+] 显示较少 [-]