Effect of Database Length and Weights on Transactions Evidence Timber Appraisal Models
Although transaction evidence appraisal (TEA) is used extensively for timber appraisal throughout the West, the effect of database length and weighing of data has been largely ignored. This paper investigates how 1-, 2-, and 3-yr model-building database lengths, with equal, biannual, quarterly, and monthly weighing schemes effect TEA prediction accuracy and responsiveness. Accuracy was measured by the "average absolute error" of predicted stumpage value from actual stumpage value. Visual inspection of the prediction trajectory served to evaluate responsiveness to market changes. Results indicate that quarterly and monthly weighing and 1-yr database length improved statistical prediction accuracy, and 1-yr database length and monthly weighing proved most responsive. West. J. Appl. For. 9(2): 71-76.
No Supplementary Data
No Article Media
Document Type: Journal Article
Affiliations: Intermountain Research Station, USDA Forest Service, Missoula, MT 59807
Publication date: 01 July 1994