You are here

Use of machine learning algorithms to predict the incidence of lead exposure in golden eagles

TitleUse of machine learning algorithms to predict the incidence of lead exposure in golden eagles
Publication TypeJournal Article
Year of Publication2009
AuthorsCraig, EH, Craig, TH, Huettmann, F, Fuller, MR
JournalPeregrine Fund Conference Proceedings 2008 Lead Conference
Abstract

Conference proceedings from 2008 Lead Conference by the Peregrine Fund https://www.peregrinefund.org/lead\textit{conference/2008PbConf}Proceedings.htmABSTRACT.–-Quantitative models can be used to predict the occurrence of wildlife relative to certain environmental conditions. Resolving the impacts of environmental contaminants on wildlife often involves complex data sets suitable for analysis with quantitative models; yet despite their potential, such models arenot commonly used. In this paper, we use data collected from wintering Golden Eagles (Aquila chrysaetos)and GIS-based models to demonstrate the use of stochastic gradient boosting, a machine learning algorithm,to examine factors most likely to influence the incidence of elevated blood lead levels. This fast, data-mining algorithm is capable of constructing predictive but sensitive and generalized models from complex contaminants datasets, and preliminary results suggest it accurately identified patterns that clarified and extended results of analyses performed using traditional statistical techniques. The managementimplications of using these models are far-reaching in their potential for identifying members of a population most at risk to contaminants, factors most likely to influence the incidence of lead contamination in a population, and potential sources of lead in the landscape. Received 1 July 2008, accepted 4 December 2008.

Username Tag: