Toward creating simpler hydrological models: A LASSO subset selection approach

Bardsley, Earl; Vetrova, Varvara; Liu, S.

doi:10.1016/j.envsoft.2015.06.008

Cited by 16 publications

(10 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The output includes coefficients for the sediment parameters, the y intercept, and a deviance ratio, which is the fraction of (null) deviance explained (equivalent to R 2 ; Friedman et al ). The elastic net approach may drop predictor variables from the model in cases where they do not significantly explain the response, consistent with least absolute shrinkage and selection operator (LASSO) regression (Bardsley et al ). Model coefficients with the largest absolute values indicate parameters with the strongest influence on the response variable.…”

Section: Methodsmentioning

confidence: 57%

Macroinvertebrate sensitivity thresholds for sediment in Virginia streams

Govenor

Krometis

Willis

et al. 2018

Integr Envir Assess & Manag

View full text Add to dashboard Cite

Sediment is the most commonly identified pollutant associated with macroinvertebrate community impairments in freshwater streams nationwide. Management of this physical stressor is complicated by the multiple measures of sediment available (e.g., suspended, dissolved, bedded) and the variability in natural "healthy" sediment loadings across ecoregions. Here we examine the relative importance of 9 sediment parameters on macroinvertebrate community health as measured by the Virginia Stream Condition Index (VSCI) across 5 ecoregions. In combination, sediment parameters explained 27.4% of variance in the VSCI in a multiregion data set and from 20.2% to 76.4% of variance for individual ecoregions. Bedded sediment parameters had a stronger influence on VSCI than did dissolved or suspended parameters in the multiregion assessment. However, assessments of individual ecoregions revealed conductivity had a key influence on VSCI in the Central Appalachian, Northern Piedmont and Piedmont ecoregions. In no case was a single sediment parameter sufficient to predict VSCI scores or individual biological metrics. Given the identification of embeddedness and conductivity as key parameters for predicting biological condition, we developed family-level sensitivity thresholds for these parameters, based on extirpation. Resulting thresholds for embeddedness were 68% for combined ecoregions, 65% for the Mountain bioregion (composed of Central Appalachian, Ridge and Valley, and Blue Ridge ecoregions), and 88% for the Piedmont bioregion (composed of Northern Piedmont and Piedmont ecoregions). Thresholds for conductivity were 366 μS/cm for combined ecoregions, 391 μS/cm for the Mountain bioregion, and 136 μS/cm for the Piedmont bioregion. These thresholds may help water quality professionals identify impaired and at-risk waters designated to support aquatic life and develop regional strategies to manage sediment-impaired streams. Inclusion of embeddedness as a restoration endpoint may be warranted; this could be facilitated by application of more quantitative, less time-intensive measurement approaches. We encourage refinement of thresholds as additional data and genus-based metrics become available. Integr Environ Assess Manag 2018;00:000-000. Published 2018. This article has been contributed to by US Government employees and their work is in the public domain in the USA.

show abstract

Section: Methodsmentioning

confidence: 57%

Macroinvertebrate sensitivity thresholds for sediment in Virginia streams

Govenor

Krometis

Willis

et al. 2018

Integr Envir Assess & Manag

View full text Add to dashboard Cite

show abstract

“…The Lasso is now very popular in a number of fields, and is integrated as part of many statistical procedures commonly used in various environmental fields (e.g. Bardsley et al, 2015).…”

Section: The Lassomentioning

confidence: 99%

EMD-regression for modelling multi-scale relationships, and application to weather-related cardiovascular mortality

Masselot

Chebana

Bélanger

et al. 2018

Science of The Total Environment

View full text Add to dashboard Cite

In a number of environmental studies, relationships between nat4ural processes are often assessed through regression analyses, using time series data. Such data are often multi-scale and non-stationary, leading to a poor accuracy of the resulting regression models and therefore to results with moderate reliability. To deal with this issue, the present paper introduces the EMD-regression methodology consisting in applying the empirical mode decomposition (EMD) algorithm on data series and then using the resulting components in regression models. The proposed methodology presents a number of advantages. First, it accounts of the issues of non-stationarity associated to the data series. Second, this approach acts as a scan for the relationship between a response variable and the predictors at different time scales, providing new insights about this relationship. To illustrate the proposed methodology it is applied to study the relationship between weather and cardiovascular mortality in Montreal, Canada. The results shed new knowledge concerning the studied relationship. For instance, they show that the humidity can cause excess mortality at the monthly time scale, which is a scale not visible in classical models. A comparison is also conducted with state of the art methods which are the generalized additive models and distributed lag models, both widely used in weather-related health studies. The comparison shows that EMD-regression achieves better prediction performances and provides more details than classical models concerning the relationship.

show abstract

“…A set of predictors can thus be divided into four disjoint parts on the basis of their relevance and redundancy ( Figure 1): strongly relevant, weakly relevant but non redundant, weakly relevant and redundant, irrelevant predictors. Since dierent partitions of weakly relevant predictors can result during the selection process, dierent quasi equally informative subsets can potentially be identied (Liu et al, 2015). The identication of these subsets requires specic metrics of relevance and redundancy and a search algorithm, especially when the dimension of the observational dataset precludes an exhaustive search in the predictors space.…”

Section: Quasi Equivalence Of Subsetsmentioning

confidence: 99%

“…On the other hand, wrappers explore the search space in a more exhaustive way and are thus likely to identify more accurate models. Variable selection algorithms are used in a variety of water resources modelling problems, such as ood regionalization (Wan Jaafar et al, 2011), statistical downscaling (Phatak et al, 2011), streamow and water quality modelling (Bardsley et al, 2015;Li et al, 2015;Creaco et al, 2016), medium-term hydro-climatic forecasts (Sharma, 2000;Noori et al, 2011) and forecast of urban water demand (Quilty et al, 2016) (see Galelli et al (2014) for a review). Whilst all these modelling problems are characterized by the presence of alternate subsets of predictors, the majority of variable selection algorithms yields one subset of predictors, thereby providing a narrow view of this source of uncertainty.…”

Section: Introductionmentioning

confidence: 99%

An information theoretic approach to select alternate subsets of predictors for data-driven hydrological models

Taormina

Galelli

Karakaya

et al. 2016

Journal of Hydrology

View full text Add to dashboard Cite

This work investigates the uncertainty associated to the presence of multiple subsets of predictors yielding data-driven models with the same, or similar, predictive accuracy. To handle this uncertainty eectively, we introduce a novel input variable selection algorithm, called Wrapper for Quasi Equally Informative Subset Selection (W-QEISS), specically conceived to identify all alternate subsets of predictors in a given dataset. The search process is based on a four-objective optimisation problem that minimises the number of selected predictors, maximises the predictive accuracy of a data-driven model and optimises two information theoretic metrics of relevance and redundancy, which guarantee that the selected subsets are highly informative and with little intra-subset similarity. The algorithm is rst tested on two synthetic test problems and then demonstrated on a real-world streamow prediction problem in the Yampa River catchment (US). Results show that complex hydro-meteorological datasets are characterised by a large number of alternate subsets of predictors, which provides useful insights on the un

show abstract

Toward creating simpler hydrological models: A LASSO subset selection approach

Cited by 16 publications

References 59 publications

Macroinvertebrate sensitivity thresholds for sediment in Virginia streams

Macroinvertebrate sensitivity thresholds for sediment in Virginia streams

EMD-regression for modelling multi-scale relationships, and application to weather-related cardiovascular mortality

An information theoretic approach to select alternate subsets of predictors for data-driven hydrological models

Contact Info

Product

Resources

About