Statistical comparison of additive regression tree methods on ecological grassland data
作者： Emily Plant , Rachel King , Jarrod Kath
DOI: 10.1016/J.ECOINF.2020.101198
关键词:
摘要: Abstract Additive tree methods are widely used in ecology. To date most ecologists have boosted regression (BRT) methods. However, Bayesian additive (BART) models may offer advantages to previously unexamined. Here we test whether BART has some benefits over the BRT method. do this use two grassland data and 13 hydroclimatic land predictor variables. The dataset contained from a period of drought as well during recovery phase after drought. response variable was trend Enhanced Vegetation Index (EVI), which is an remotely sensed indicator degradation recovery. settable parameters both (BRT BART) were varied compare performance each evaluated using three prediction error statistics; root mean square (RMSE), absolute (MAE), coefficient determination (R2). best across assessed by inspecting relative importance variables statistics. exhibited similar selection abilities, but method generated with or more favourable statistics than (BART explained additional 10.17% 11.92% variation models). Our results indicate that BARTs be effective at modelling ecological BRTs. also had shorter run times, reasonable defaults its software implementation, greater functionality said beyond model building functions. Ecologists approaches benefit suggest their alongside commonly studies.
usq.edu.au
sci-hub.st 参考文章(33)
Matthew Wiener, Andy Liaw, Classification and Regression by randomForest ,(2007)
Robert Tibshirani, Trevor Hastie, Daniela Witten, Gareth James, An Introduction to Statistical Learning: With Applications in R ,(2013)
Jesper Erenskjold Moeslund, Lars Arge, Peder Klith Bøcher, Tommy Dalgaard, Rasmus Ejrnæs, Mette Vestergaard Odgaard, Jens-Christian Svenning, Topographically controlled soil moisture drives plant diversity patterns within grasslands Biodiversity and Conservation. ,vol. 22, pp. 2151- 2166 ,(2013) , 10.1007/S10531-013-0442-3
Yoav Freund, Robert E Schapire, A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting conference on learning theory. ,vol. 55, pp. 119- 139 ,(1997) , 10.1006/JCSS.1997.1504
Carlos P Carmona, Achim Röder, Francisco M Azcárate, Begoña Peco, None, Grazing management or physiography? Factors controlling vegetation recovery in Mediterranean grasslands Ecological Modelling. ,vol. 251, pp. 73- 84 ,(2013) , 10.1016/J.ECOLMODEL.2012.12.005
Weidong Zhang, Shufen Yuan, Ning Hu, Yilai Lou, Silong Wang, Predicting soil fauna effect on plant litter decomposition by using boosted regression trees. Soil Biology & Biochemistry. ,vol. 82, pp. 81- 86 ,(2015) , 10.1016/J.SOILBIO.2014.12.016
CJ Willmott, K Matsuura, Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance Climate Research. ,vol. 30, pp. 79- 82 ,(2005) , 10.3354/CR030079
Christian Kampichler, Ralf Wieland, Sophie Calmé, Holger Weissenberger, Stefan Arriaga-Weiss, Classification in conservation biology: A comparison of five machine-learning methods Ecological Informatics. ,vol. 5, pp. 441- 450 ,(2010) , 10.1016/J.ECOINF.2010.06.003
Y. Freund, Boosting a weak learning algorithm by majority Information & Computation. ,vol. 121, pp. 256- 285 ,(1995) , 10.1006/INCO.1995.1136
S. C. Cunningham, R. Mac Nally, J. Read, P. J. Baker, M. White, J. R. Thomson, P. Griffioen, A Robust Technique for Mapping Vegetation Condition Across a Major River System Ecosystems. ,vol. 12, pp. 207- 219 ,(2009) , 10.1007/S10021-008-9218-0