Details
| Original language | English |
|---|---|
| Pages (from-to) | 245-272 |
| Number of pages | 28 |
| Journal | Environmental and Ecological Statistics |
| Volume | 31 |
| Issue number | 2 |
| Early online date | 1 Feb 2024 |
| Publication status | Published - Jun 2024 |
Abstract
This study presents a comparative analysis of three predictive models with an increasing degree of flexibility: hidden dynamic geostatistical models (HDGM), generalised additive mixed models (GAMM), and the random forest spatiotemporal kriging models (RFSTK). These models are evaluated for their effectiveness in predicting PM 2.5 concentrations in Lombardy (North Italy) from 2016 to 2020. Despite differing methodologies, all models demonstrate proficient capture of spatiotemporal patterns within air pollution data with similar out-of-sample performance. Furthermore, the study delves into station-specific analyses, revealing variable model performance contingent on localised conditions. Model interpretation, facilitated by parametric coefficient analysis and partial dependence plots, unveils consistent associations between predictor variables and PM 2.5 concentrations. Despite nuanced variations in modelling spatiotemporal correlations, all models effectively accounted for the underlying dependence. In summary, this study underscores the efficacy of conventional techniques in modelling correlated spatiotemporal data, concurrently highlighting the complementary potential of Machine Learning and classical statistical approaches.
Keywords
- Air pollution, Generalised additive mixed model, Geostatistics, Hidden dynamic geostatistical model, Machine learning, Random forest spatiotemporal kriging, Spatiotemporal process
ASJC Scopus subject areas
- Mathematics(all)
- Statistics and Probability
- Environmental Science(all)
- General Environmental Science
- Decision Sciences(all)
- Statistics, Probability and Uncertainty
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
In: Environmental and Ecological Statistics, Vol. 31, No. 2, 06.2024, p. 245-272.
Research output: Contribution to journal › Article › Research › peer review
}
TY - JOUR
T1 - Spatiotemporal modelling of PM 2.5 concentrations in Lombardy (Italy)
T2 - a comparative study
AU - Otto, Philipp
AU - Fusta Moro, Alessandro
AU - Rodeschini, Jacopo
AU - Shaboviq, Qendrim
AU - Ignaccolo, Rosaria
AU - Golini, Natalia
AU - Cameletti, Michela
AU - Maranzano, Paolo
AU - Finazzi, Francesco
AU - Fassò, Alessandro
N1 - Publisher Copyright: © The Author(s) 2024.
PY - 2024/6
Y1 - 2024/6
N2 - This study presents a comparative analysis of three predictive models with an increasing degree of flexibility: hidden dynamic geostatistical models (HDGM), generalised additive mixed models (GAMM), and the random forest spatiotemporal kriging models (RFSTK). These models are evaluated for their effectiveness in predicting PM 2.5 concentrations in Lombardy (North Italy) from 2016 to 2020. Despite differing methodologies, all models demonstrate proficient capture of spatiotemporal patterns within air pollution data with similar out-of-sample performance. Furthermore, the study delves into station-specific analyses, revealing variable model performance contingent on localised conditions. Model interpretation, facilitated by parametric coefficient analysis and partial dependence plots, unveils consistent associations between predictor variables and PM 2.5 concentrations. Despite nuanced variations in modelling spatiotemporal correlations, all models effectively accounted for the underlying dependence. In summary, this study underscores the efficacy of conventional techniques in modelling correlated spatiotemporal data, concurrently highlighting the complementary potential of Machine Learning and classical statistical approaches.
AB - This study presents a comparative analysis of three predictive models with an increasing degree of flexibility: hidden dynamic geostatistical models (HDGM), generalised additive mixed models (GAMM), and the random forest spatiotemporal kriging models (RFSTK). These models are evaluated for their effectiveness in predicting PM 2.5 concentrations in Lombardy (North Italy) from 2016 to 2020. Despite differing methodologies, all models demonstrate proficient capture of spatiotemporal patterns within air pollution data with similar out-of-sample performance. Furthermore, the study delves into station-specific analyses, revealing variable model performance contingent on localised conditions. Model interpretation, facilitated by parametric coefficient analysis and partial dependence plots, unveils consistent associations between predictor variables and PM 2.5 concentrations. Despite nuanced variations in modelling spatiotemporal correlations, all models effectively accounted for the underlying dependence. In summary, this study underscores the efficacy of conventional techniques in modelling correlated spatiotemporal data, concurrently highlighting the complementary potential of Machine Learning and classical statistical approaches.
KW - Air pollution
KW - Generalised additive mixed model
KW - Geostatistics
KW - Hidden dynamic geostatistical model
KW - Machine learning
KW - Random forest spatiotemporal kriging
KW - Spatiotemporal process
UR - http://www.scopus.com/inward/record.url?scp=85183928401&partnerID=8YFLogxK
U2 - 10.1007/s10651-023-00589-0
DO - 10.1007/s10651-023-00589-0
M3 - Article
AN - SCOPUS:85183928401
VL - 31
SP - 245
EP - 272
JO - Environmental and Ecological Statistics
JF - Environmental and Ecological Statistics
SN - 1352-8505
IS - 2
ER -