Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation
Importance of target-oriented validation strategies for spatio-temporal prediction models is illustrated using two case studies: (1) modelling of air temperature () in Antarctica, and (2) modelling of volumetric water content (VW) for the R.J. Cook Agronomy Farm, USA. Performance of a random k-fold...
Published in: | Environmental Modelling & Software |
---|---|
Main Authors: | , , , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
Elsevier
2018
|
Subjects: | |
Online Access: | https://oceanrep.geomar.de/id/eprint/50207/ https://oceanrep.geomar.de/id/eprint/50207/1/Meyer.pdf https://doi.org/10.1016/j.envsoft.2017.12.001 |
id |
ftoceanrep:oai:oceanrep.geomar.de:50207 |
---|---|
record_format |
openpolar |
spelling |
ftoceanrep:oai:oceanrep.geomar.de:50207 2023-05-15T13:37:37+02:00 Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation Meyer, Hanna Reudenbach, Christoph Hengl, Tomislav Katurji, Marwan Nauss, Thomas 2018-03 text https://oceanrep.geomar.de/id/eprint/50207/ https://oceanrep.geomar.de/id/eprint/50207/1/Meyer.pdf https://doi.org/10.1016/j.envsoft.2017.12.001 en eng Elsevier https://oceanrep.geomar.de/id/eprint/50207/1/Meyer.pdf Meyer, H., Reudenbach, C., Hengl, T., Katurji, M. and Nauss, T. (2018) Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation. Environmental Modelling & Software, 101 . pp. 1-9. DOI 10.1016/j.envsoft.2017.12.001 <https://doi.org/10.1016/j.envsoft.2017.12.001>. doi:10.1016/j.envsoft.2017.12.001 info:eu-repo/semantics/restrictedAccess Article PeerReviewed 2018 ftoceanrep https://doi.org/10.1016/j.envsoft.2017.12.001 2023-04-07T15:51:17Z Importance of target-oriented validation strategies for spatio-temporal prediction models is illustrated using two case studies: (1) modelling of air temperature () in Antarctica, and (2) modelling of volumetric water content (VW) for the R.J. Cook Agronomy Farm, USA. Performance of a random k-fold cross-validation (CV) was compared to three target-oriented strategies: Leave-Location-Out (LLO), Leave-Time-Out (LTO), and Leave-Location-and-Time-Out (LLTO) CV. Results indicate that considerable differences between random k-fold ( = 0.9 for and 0.92 for VW) and target-oriented CV (LLO = 0.24 for and 0.49 for VW) exist, highlighting the need for target-oriented validation to avoid an overoptimistic view on models. Differences between random k-fold and target-oriented CV indicate spatial over-fitting caused by misleading variables. To decrease over-fitting, a forward feature selection in conjunction with target-oriented CV is proposed. It decreased over-fitting and simultaneously improved target-oriented performances (LLO CV = 0.47 for and 0.55 for VW). Article in Journal/Newspaper Antarc* Antarctica OceanRep (GEOMAR Helmholtz Centre für Ocean Research Kiel) Environmental Modelling & Software 101 1 9 |
institution |
Open Polar |
collection |
OceanRep (GEOMAR Helmholtz Centre für Ocean Research Kiel) |
op_collection_id |
ftoceanrep |
language |
English |
description |
Importance of target-oriented validation strategies for spatio-temporal prediction models is illustrated using two case studies: (1) modelling of air temperature () in Antarctica, and (2) modelling of volumetric water content (VW) for the R.J. Cook Agronomy Farm, USA. Performance of a random k-fold cross-validation (CV) was compared to three target-oriented strategies: Leave-Location-Out (LLO), Leave-Time-Out (LTO), and Leave-Location-and-Time-Out (LLTO) CV. Results indicate that considerable differences between random k-fold ( = 0.9 for and 0.92 for VW) and target-oriented CV (LLO = 0.24 for and 0.49 for VW) exist, highlighting the need for target-oriented validation to avoid an overoptimistic view on models. Differences between random k-fold and target-oriented CV indicate spatial over-fitting caused by misleading variables. To decrease over-fitting, a forward feature selection in conjunction with target-oriented CV is proposed. It decreased over-fitting and simultaneously improved target-oriented performances (LLO CV = 0.47 for and 0.55 for VW). |
format |
Article in Journal/Newspaper |
author |
Meyer, Hanna Reudenbach, Christoph Hengl, Tomislav Katurji, Marwan Nauss, Thomas |
spellingShingle |
Meyer, Hanna Reudenbach, Christoph Hengl, Tomislav Katurji, Marwan Nauss, Thomas Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation |
author_facet |
Meyer, Hanna Reudenbach, Christoph Hengl, Tomislav Katurji, Marwan Nauss, Thomas |
author_sort |
Meyer, Hanna |
title |
Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation |
title_short |
Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation |
title_full |
Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation |
title_fullStr |
Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation |
title_full_unstemmed |
Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation |
title_sort |
improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation |
publisher |
Elsevier |
publishDate |
2018 |
url |
https://oceanrep.geomar.de/id/eprint/50207/ https://oceanrep.geomar.de/id/eprint/50207/1/Meyer.pdf https://doi.org/10.1016/j.envsoft.2017.12.001 |
genre |
Antarc* Antarctica |
genre_facet |
Antarc* Antarctica |
op_relation |
https://oceanrep.geomar.de/id/eprint/50207/1/Meyer.pdf Meyer, H., Reudenbach, C., Hengl, T., Katurji, M. and Nauss, T. (2018) Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation. Environmental Modelling & Software, 101 . pp. 1-9. DOI 10.1016/j.envsoft.2017.12.001 <https://doi.org/10.1016/j.envsoft.2017.12.001>. doi:10.1016/j.envsoft.2017.12.001 |
op_rights |
info:eu-repo/semantics/restrictedAccess |
op_doi |
https://doi.org/10.1016/j.envsoft.2017.12.001 |
container_title |
Environmental Modelling & Software |
container_volume |
101 |
container_start_page |
1 |
op_container_end_page |
9 |
_version_ |
1766095053622607872 |