Dataset Information

A New Technique for Evaluating Land-use Regression Models and Their Impact on Health Effect Estimates.

ABSTRACT: BACKGROUND:Leave-one-out cross-validation that fails to account for variable selection does not properly reflect prediction accuracy when the number of training sites is small. The impact on health effect estimates has rarely been studied. The objective of this study was to develop an improved validation procedure for land-use regression models with variable selection and investigate health effect estimates in relation to land-use regression model performance. METHODS:We randomly generated 10 training and test sets for nitrogen dioxide and particulate matter. For each training set, we developed models and evaluated them using a cross-holdout validation approach. Cross-holdout validation develops new models for each evaluation compared with refitting the model without variable selection, as in standard leave-one-out cross-validation. We also implemented holdout validation, which evaluates model predictions using independent test sets. We evaluated the relationship between cross-holdout validation and holdout validation R and estimates of the association between air pollution and forced vital capacity in the Dutch birth cohort. RESULTS:Cross-holdout validation Rs were generally identical to holdout validation Rs, but were notably smaller than leave-one-out cross-validation Rs. Decreases in forced vital capacity in relation to air pollution exposure were larger for land-use regression models that had larger holdout validation and cross-holdout validation Rs rather than leave-one-out cross-validation R. CONCLUSION:Cross-holdout validation accurately reflects predictive ability of land-use regression models and is a useful validation approach for small datasets. Land-use regression predictive ability in terms of holdout validation and cross-holdout validation rather than leave-one-out cross-validation was associated with the magnitude of health effect estimates in a case study.

SUBMITTER: Wang M

PROVIDER: S-EPMC5221608 | biostudies-literature | 2016 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A New Technique for Evaluating Land-use Regression Models and Their Impact on Health Effect Estimates.

Wang Meng M Brunekreef Bert B Gehring Ulrike U Szpiro Adam A Hoek Gerard G Beelen Rob R

Epidemiology (Cambridge, Mass.) 20160101 1

<h4>Background</h4>Leave-one-out cross-validation that fails to account for variable selection does not properly reflect prediction accuracy when the number of training sites is small. The impact on health effect estimates has rarely been studied. The objective of this study was to develop an improved validation procedure for land-use regression models with variable selection and investigate health effect estimates in relation to land-use regression model performance.<h4>Methods</h4>We randomly ...[more]

PMID: 26426941

Similar Datasets

Project description:Land-use regression (LUR) models have been developed to estimate spatial distributions of traffic-related pollutants. Several studies have examined spatial autocorrelation among residuals in LUR models, but few utilized spatial residual information in model prediction, or examined the impact of modeling methods, monitoring site selection, or traffic data quality on LUR performance. This study aims to improve spatial models for traffic-related pollutants using generalized additive models (GAM) combined with cokriging of spatial residuals. Specifically, we developed spatial models for nitrogen dioxide (NO(2)) and nitrogen oxides (NO(x)) concentrations in Southern California separately for two seasons (summer and winter) based on over 240 sampling locations. Pollutant concentrations were disaggregated into three components: local means, spatial residuals, and normal random residuals. Local means were modeled by GAM. Spatial residuals were cokriged with global residuals at nearby sampling locations that were spatially auto-correlated. We compared this two-stage approach with four commonly-used spatial models: universal kriging, multiple linear LUR and GAM with and without a spatial smoothing term. Leave-one-out cross validation was conducted for model validation and comparison purposes. The results show that our GAM plus cokriging models predicted summer and winter NO(2) and NO(x) concentration surfaces well, with cross validation R(2) values ranging from 0.88 to 0.92. While local covariates accounted for partial variance of the measured NO(2) and NO(x) concentrations, spatial autocorrelation accounted for about 20% of the variance. Our spatial GAM model improved R(2) considerably compared to the other four approaches. Conclusively, our two-stage model captured summer and winter differences in NO(2) and NO(x) spatial distributions in Southern California well. When sampling location selection cannot be optimized for the intended model and fewer covariates are available as predictors for the model, the two-stage model is more robust compared to multiple linear regression models.

Project description:Land-use change is one of the biggest threats to biodiversity globally. The effects of land use on biodiversity manifest primarily at local scales which are not captured by the coarse spatial grain of current global land-use mapping. Assessments of land-use impacts on biodiversity across large spatial extents require data at a similar spatial grain to the ecological processes they are assessing. Here, we develop a method for statistically downscaling mapped land-use data that combines generalized additive modeling and constrained optimization. This method was applied to the 0.5° Land-use Harmonization data for the year 2005 to produce global 30″ (approx. 1 km(2)) estimates of five land-use classes: primary habitat, secondary habitat, cropland, pasture, and urban. The original dataset was partitioned into 61 bio-realms (unique combinations of biome and biogeographical realm) and downscaled using relationships with fine-grained climate, land cover, landform, and anthropogenic influence layers. The downscaled land-use data were validated using the PREDICTS database and the geoWiki global cropland dataset. Application of the new method to all 61 bio-realms produced global fine-grained layers from the 2005 time step of the Land-use Harmonization dataset. Coarse-scaled proportions of land use estimated from these data compared well with those estimated in the original datasets (mean R (2): 0.68 ± 0.19). Validation with the PREDICTS database showed the new downscaled land-use layers improved discrimination of all five classes at PREDICTS sites (P < 0.0001 in all cases). Additional validation of the downscaled cropping layer with the geoWiki layer showed an R (2) improvement of 0.12 compared with the Land-use Harmonization data. The downscaling method presented here produced the first global land-use dataset at a spatial grain relevant to ecological processes that drive changes in biodiversity over space and time. Integrating these data with biodiversity measures will enable the reporting of land-use impacts on biodiversity at a finer resolution than previously possible. Furthermore, the general method presented here could be useful to others wishing to downscale similarly constrained coarse-resolution data for other environmental variables.

Dataset Information

A New Technique for Evaluating Land-use Regression Models and Their Impact on Health Effect Estimates.

Publications

A New Technique for Evaluating Land-use Regression Models and Their Impact on Health Effect Estimates.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets