Using ROC curves to compare neural networks and logistic regression for modeling individual noncatastrophic tree mortality

Abstract

The performance of two classifiers, logistic regression and neural networks, are compared for modeling noncatastrophic individual tree mortality for 21 species of trees in West Virginia. The output of the classifier is usually a continuous number between 0 and 1. A threshold is selected between 0 and 1 and all of the trees below the threshold are classified as mortality trees and all of the trees above the threshold are classified as survival trees. Selecting the threshold that has both a high sensitivity and specificity is a major decision. A receiver operating characteristic (ROC) curve graphically describes the performance of the classifier without the requirement of a threshold. Its accuracy is measured by the area under the curve (AUC). A neural network is the superior classifier because it has a higher AUC statistic.

Parent Publication

• 13th Central Hardwoods Forest Conference

Citation

King, Susan L. 2003. Using ROC curves to compare neural networks and logistic regression for modeling individual noncatastrophic tree mortality. In: Van Sambeek, J. W.; Dawson, Jeffery O.; Ponder Jr., Felix; Loewenstein, Edward F.; Fralish, James S., eds. Proceedings of the 13th Central Hardwood Forest Conference; Gen. Tech. Rep. NC-234. St. Paul, MN: U.S. Department of Agriculture, Forest Service, North Central Research Station: 349-358