Random forest as a generic framework for predictive modeling of spatial and spatio-temporal variables

Hengl, Tomislav; Nussbaum, Madlene; Wright, Marvin N.; Heuvelink, Gerard B.M.; Gräler, Benedikt (2018). Random forest as a generic framework for predictive modeling of spatial and spatio-temporal variables PeerJ, 6, e5518. 10.7717/peerj.5518

Preview

Text 5518.pdf - Published Version Available under License Publisher holds Copyright. Download (7MB) | Preview

Random forest and similar Machine Learning techniques are already used to generate spatial predictions, but spatial location of points (geography) is often ignored in the modeling process. Spatial auto-correlation, especially if still existent in the cross-validation residuals, indicates that the predictions are maybe biased, and this is suboptimal. This paper presents a random forest for spatial predictions framework (RFsp) where buffer distances from observation points are used as explanatory variables, thus incorporating geographical proximity effects into the prediction process. The RFsp framework is illustrated with examples that use textbook datasets and apply spatial and spatio-temporal prediction to numeric, binary, categorical, multivariate and spatiotemporal variables. Performance of the RFsp framework is compared with the state-of-the-art kriging techniques using fivefold cross-validation with refitting. The results show that RFsp can obtain equally accurate and unbiased predictions as different versions of kriging. Advantages of using RFsp over kriging are that it needs no rigid statistical assumptions about the distribution and stationarity of the target variable, it is more flexible towards incorporating, combining and extending covariates of different types, and it possibly yields more informative maps characterizing the prediction error. RFsp appears to be especially attractive for building multivariate spatial prediction models that can be used as “knowledge engines” in various geoscience fields. Some disadvantages of RFsp are the exponentially growing computational intensity with increase of calibration data and covariates and the high sensitivity of predictions to input data quality. The key to the success of the RFsp framework might be the training data quality—especially quality of spatial sampling (to minimize extrapolation problems and any type of bias in data), and quality of model validation (to ensure that accuracy is not effected by overfitting). For many data sets, especially those with lower number of points and covariates and close-to-linear relationships, model-based geostatistics can still lead to more accurate predictions than RFsp.

Item Type:	Journal Article (Original Article)
Division/Institute:	School of Agricultural, Forest and Food Sciences HAFL > Resource-efficient agricultural production systems School of Agricultural, Forest and Food Sciences HAFL > Agriculture
Name:	Hengl, Tomislav; Nussbaum, Madlene0000-0002-6808-8956; Wright, Marvin N.; Heuvelink, Gerard B.M. and Gräler, Benedikt
Subjects:	Q Science > Q Science (General) S Agriculture > S Agriculture (General) T Technology > T Technology (General)
ISSN:	2167-8359
Language:	English
Submitter:	David Zimmer
Date Deposited:	12 Feb 2020 08:34
Last Modified:	15 Apr 2021 21:45
Publisher DOI:	10.7717/peerj.5518
PubMed ID:	30186691
Uncontrolled Keywords:	Biogeography; Soil Science; Computational Science; Data Mining and Machine Learning; Spatial and Geographic; Information Science
ARBOR DOI:	10.24451/arbor.8184
URI:	https://arbor.bfh.ch/id/eprint/8184