A framework for constructing machine learning models with feature set optimisation for evapotranspiration partitioning
A deeper understanding of the drivers of evapotranspiration and the modelling of its constituent parts (evaporation and transpiration) may be of significant importance to the monitoring and management of water resources globally over the coming decades. In this work a framework was developed to iden...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2022-12-01
|
Series: | Applied Computing and Geosciences |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S2590197422000271 |
_version_ | 1811190733982400512 |
---|---|
author | Adam Stapleton Elke Eichelmann Mark Roantree |
author_facet | Adam Stapleton Elke Eichelmann Mark Roantree |
author_sort | Adam Stapleton |
collection | DOAJ |
description | A deeper understanding of the drivers of evapotranspiration and the modelling of its constituent parts (evaporation and transpiration) may be of significant importance to the monitoring and management of water resources globally over the coming decades. In this work a framework was developed to identify the best performing machine learning algorithm from a candidate set, select optimal predictive features and rank features in terms of their importance to predictive accuracy. The experiments conducted in this work used 3 separate feature sets across 4 wetland sites as input into 8 candidate machine learning algorithms, providing 96 sets of experimental configurations. Given this high number of parameters, our results show strong evidence that there is no singularly optimal machine learning algorithm or feature set across all of the wetland sites studied despite their similarities. At each of the sites at least one model was identified that improved on the predictive performance of our baseline. A key finding discovered when examining feature importance is that methane flux, a feature whose relationship with evapotranspiration is not generally examined, may contribute to further biophysical process understanding. This work demonstrates the applicability of a machine learning framework for evapotranspiration partitioning that is independent of domain knowledge, producing improved models for partitioning and identifying new and useful predictive features. |
first_indexed | 2024-04-11T14:55:13Z |
format | Article |
id | doaj.art-abd24322fae5492aac27c586563f3e05 |
institution | Directory Open Access Journal |
issn | 2590-1974 |
language | English |
last_indexed | 2024-04-11T14:55:13Z |
publishDate | 2022-12-01 |
publisher | Elsevier |
record_format | Article |
series | Applied Computing and Geosciences |
spelling | doaj.art-abd24322fae5492aac27c586563f3e052022-12-22T04:17:17ZengElsevierApplied Computing and Geosciences2590-19742022-12-0116100105A framework for constructing machine learning models with feature set optimisation for evapotranspiration partitioningAdam Stapleton0Elke Eichelmann1Mark Roantree2School of Computing, Dublin City University, Dublin 9, Ireland; Corresponding author.School of Biology and Environmental Science, University College Dublin, Dublin 4, IrelandInsight Centre for Data Analytics, Dublin City University, Dublin 9, IrelandA deeper understanding of the drivers of evapotranspiration and the modelling of its constituent parts (evaporation and transpiration) may be of significant importance to the monitoring and management of water resources globally over the coming decades. In this work a framework was developed to identify the best performing machine learning algorithm from a candidate set, select optimal predictive features and rank features in terms of their importance to predictive accuracy. The experiments conducted in this work used 3 separate feature sets across 4 wetland sites as input into 8 candidate machine learning algorithms, providing 96 sets of experimental configurations. Given this high number of parameters, our results show strong evidence that there is no singularly optimal machine learning algorithm or feature set across all of the wetland sites studied despite their similarities. At each of the sites at least one model was identified that improved on the predictive performance of our baseline. A key finding discovered when examining feature importance is that methane flux, a feature whose relationship with evapotranspiration is not generally examined, may contribute to further biophysical process understanding. This work demonstrates the applicability of a machine learning framework for evapotranspiration partitioning that is independent of domain knowledge, producing improved models for partitioning and identifying new and useful predictive features.http://www.sciencedirect.com/science/article/pii/S2590197422000271Machine learningEvapotranspirationOptimisationAtmospheric sciencesGeophysicsEnvironmental sciences |
spellingShingle | Adam Stapleton Elke Eichelmann Mark Roantree A framework for constructing machine learning models with feature set optimisation for evapotranspiration partitioning Applied Computing and Geosciences Machine learning Evapotranspiration Optimisation Atmospheric sciences Geophysics Environmental sciences |
title | A framework for constructing machine learning models with feature set optimisation for evapotranspiration partitioning |
title_full | A framework for constructing machine learning models with feature set optimisation for evapotranspiration partitioning |
title_fullStr | A framework for constructing machine learning models with feature set optimisation for evapotranspiration partitioning |
title_full_unstemmed | A framework for constructing machine learning models with feature set optimisation for evapotranspiration partitioning |
title_short | A framework for constructing machine learning models with feature set optimisation for evapotranspiration partitioning |
title_sort | framework for constructing machine learning models with feature set optimisation for evapotranspiration partitioning |
topic | Machine learning Evapotranspiration Optimisation Atmospheric sciences Geophysics Environmental sciences |
url | http://www.sciencedirect.com/science/article/pii/S2590197422000271 |
work_keys_str_mv | AT adamstapleton aframeworkforconstructingmachinelearningmodelswithfeaturesetoptimisationforevapotranspirationpartitioning AT elkeeichelmann aframeworkforconstructingmachinelearningmodelswithfeaturesetoptimisationforevapotranspirationpartitioning AT markroantree aframeworkforconstructingmachinelearningmodelswithfeaturesetoptimisationforevapotranspirationpartitioning AT adamstapleton frameworkforconstructingmachinelearningmodelswithfeaturesetoptimisationforevapotranspirationpartitioning AT elkeeichelmann frameworkforconstructingmachinelearningmodelswithfeaturesetoptimisationforevapotranspirationpartitioning AT markroantree frameworkforconstructingmachinelearningmodelswithfeaturesetoptimisationforevapotranspirationpartitioning |