Probabilistic predictive principal component analysis for spatially misaligned and high‐dimensional air pollution data with missing observations

Vu, Phuong T.; Larson, Timothy V.; Szpiro, Adam A.

doi:10.1002/env.2614

Cited by 4 publications

(14 citation statements)

References 56 publications

(110 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We also illustrate how SMC can produce reliable estimates of the entire pollutant profile at new locations while LRMC simply cannot. In addition, we show the merits of SMC under spatial misalignment typically seen in air pollution cohort studies by reproducing and comparing to simulated results and data applications given in Vu et al (2020).…”

Section: Discussionmentioning

confidence: 77%

“…Our proposed SMC model offers an alternative approach, with similar goals to ProPrPCA, but using convex optimization. Thus, in this section, we reproduce the simulations and data application in Vu et al (2020), to compare SMC directly to ProPrPCA, PredPCA, and traditional PCA.…”

Section: Predictive Performance Under Spatial Misalignment: a Compari...mentioning

confidence: 99%

“…The training data in each iteration may consist of either the remaining complete site ("complete" training data), or all remaining site ("full" training data). Similar to Vu et al (2020), we focus on the prediction for each PC instead of reconstructing the data matrix, because in practice, these PCs would represent the pollutant profile, which could be used as an effect modifier in a model investigating the associations between total mass of particulate matter and health outcomes.…”

Section: Data Applicationmentioning

confidence: 99%

“…While capable of leveraging both the LR and spatial structures of the monitoring data, in its current form PredPCA cannot handle input data with missing observations. This motivated a probabilistic version of PredPCA, ProPrPCA (Vu et al, 2020), that can handle incomplete monitoring data. Using external covariates, the unobserved data at new locations can be estimated directly from the hierarchical model assumptions, or by incorporating the LR structure of monitoring data in a separate spatial prediction model.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Spatial matrix completion for spatially misaligned and high‐dimensional air pollution data

Szpiro

Simon

2021

Environmetrics

Self Cite

View full text Add to dashboard Cite

In health‐pollution cohort studies, accurate predictions of pollutant concentrations at new locations are needed, since the locations of fixed monitoring sites and study participants are often spatially misaligned. For multi‐pollution data, principal component analysis (PCA) is often incorporated to obtain low‐rank (LR) structure of the data prior to spatial prediction. Recently developed predictive PCA modifies the traditional algorithm to improve the overall predictive performance by leveraging both LR and spatial structures within the data. However, predictive PCA requires complete data or an initial imputation step. Nonparametric imputation techniques without accounting for spatial information may distort the underlying structure of the data, and thus further reduce the predictive performance. We propose a convex optimization problem inspired by the LR matrix completion framework and develop a proximal algorithm to solve it. Missing data are imputed and handled concurrently within the algorithm, which eliminates the necessity of a separate imputation step. We review the connections among those existing methods developed for spatially misaligned multivariate data, and show that our algorithm has lower computational burden and leads to reliable predictive performance as the severity of missing data increases.

show abstract

Section: Discussionmentioning

confidence: 77%

Section: Predictive Performance Under Spatial Misalignment: a Compari...mentioning

confidence: 99%

Section: Data Applicationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Spatial matrix completion for spatially misaligned and high‐dimensional air pollution data

Szpiro

Simon

2021

Environmetrics

Self Cite

View full text Add to dashboard Cite

show abstract

“…Regarding pollutants concentrations, increasing attention is being given to latent component models; see, as an example [13] and for the problem of misalignment. In particular, the use of the INLA-SPDE approach for misalignment between pollutant concentration and epidemiological data [14] and PCA based methods with missing data [15].…”

Section: Introductionmentioning

confidence: 99%

Statistical Modeling of the Early-Stage Impact of a New Traffic Policy in Milan, Italy

Maranzano

Fassò

Pelagatti

et al. 2020

IJERPH

View full text Add to dashboard Cite

Most urban areas of the Po basin in the North of Italy are persistently affected by poor air quality and difficulty in disposing of airborne pollutants. In this context, the municipality of Milan started a multi-year progressive policy based on an extended limited traffic zone (Area B). Starting on 25 February 2019, the first phase partially restricted the circulation of some classes of highly polluting vehicles on the territory, in particular, Euro 0 petrol vehicles and Euro 0 to 3 diesel vehicles, excluding public transport. This is the early-stage of a long term policy that will restrict access to an increasing number of vehicles. The goal of this paper is to evaluate the early-stage impact of this policy on two specific vehicle-generated pollutants: total nitrogen oxides (NO x ) and nitrogen dioxide (NO 2 ), which are gathered by Lombardy Regional Agency for Environmental Protection (ARPA Lombardia). We use a statistical model for time series intervention analysis based on unobservable components. We use data from 2014 to 2018 for pre-policy model selection and the relatively short period up to September 2019 for early-stage policy assessment. We include weather conditions, socio-economic factors, and a counter-factual, given by the concentration of the same pollutant in other important neighbouring cities. Although the average concentrations reduced after the policy introduction, this paper argues that this could be due to other factors. Considering that the short time window may be not long enough for social adaptation to the new rules, our model does not provide statistical evidence of a positive policy effect for NO x and NO 2 . Instead, in one of the most central monitoring stations, a significant negative impact is found.

show abstract

Risk assessment and prevention of air pollution to protect citizen health based on statistical data: a case study of Zhengzhou, China

2021

Stoch Environ Res Risk Assess

View full text Add to dashboard Cite

Probabilistic predictive principal component analysis for spatially misaligned and high‐dimensional air pollution data with missing observations

Cited by 4 publications

References 56 publications

Spatial matrix completion for spatially misaligned and high‐dimensional air pollution data

Spatial matrix completion for spatially misaligned and high‐dimensional air pollution data

Statistical Modeling of the Early-Stage Impact of a New Traffic Policy in Milan, Italy

Risk assessment and prevention of air pollution to protect citizen health based on statistical data: a case study of Zhengzhou, China

Contact Info

Product

Resources

About