Package: pcpr 1.0.0.9000

Lawrence G. Chillrud

pcpr: Principal Component Pursuit for Environmental Epidemiology

Implementation of the pattern recognition technique Principal Component Pursuit tailored to environmental health data, as described in Gibson et al (2022) <doi:10.1289/EHP10479>.

Authors:Lawrence G. Chillrud [aut, cre, cph], Jaime Benavides [aut], Elizabeth A. Gibson [aut], Junhui Zhang [aut], Jingkai Yan [aut], John N. Wright [aut], Jeff Goldsmith [aut], Marianthi-Anna Kioumourtzoglou [aut], Columbia University [fnd]

pcpr_1.0.0.9000.tar.gz
pcpr_1.0.0.9000.zip(r-4.7-any)pcpr_1.0.0.9000.zip(r-4.6-any)pcpr_1.0.0.9000.zip(r-4.5-any)
pcpr_1.0.0.9000.tgz(r-4.6-any)pcpr_1.0.0.9000.tgz(r-4.5-any)
pcpr_1.0.0.9000.tar.gz(r-4.7-any)pcpr_1.0.0.9000.tar.gz(r-4.6-any)
pcpr_1.0.0.9000.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
pcpr/json (API)

# Install 'pcpr' in R:
install.packages('pcpr', repos = c('https://columbia-prime.r-universe.dev', 'https://cloud.r-project.org'))

Bug tracker:https://github.com/columbia-prime/pcpr/issues

Pkgdown/docs site:https://columbia-prime.github.io

Datasets:
  • queens - Daily chemical concentrations of 26 PM2.5 species from Queens, NYC

On CRAN:

Conda:

dimensionality-reductionenvironmental-healthenvironmental-mixturesepidemiologymachine-learningpattern-recognitionpublic-healthstatistical-modeling

5.28 score 7 stars 18 scripts 340 downloads 13 exports 28 dependencies

Last updated from:d4543329c8. Checks:9 OK. Indexed: yes.

TargetResultTimeFilesSyslog
linux-develOK124
source / vignettesOK220
linux-releaseOK122
macos-releaseOK183
macos-oldrelOK171
windows-develOK71
windows-releaseOK95
windows-oldrelOK78
wasm-releaseOK100

Exports:get_pcp_defaultsgrid_search_cvhard_thresholdimpute_matrixmatrix_rankproj_rank_rroot_pcprrmcsim_datasim_lodsim_nasingsparsity

Dependencies:backportscheckmateclicodetoolsdigestdplyrforeachfurrrfuturegenericsglobalsglueiteratorslifecyclelistenvmagrittrparallellypillarpkgconfigprogressrpurrrR6rlangtibbletidyselectutf8vctrswithr

Air pollution source apportionment with PCP
Preprocessing | Exploring the raw data | Data standardization | Principal component pursuit | Model selection | Grid search for parameter tuning | Running PCP | The recovered L matrix | The recovered S matrix | PCP and NMF for source apportionment | PCP-NMF Loadings | Variance Explained | Loadings replotted | PCP-NMF Scores | Conclusion | References
Rendered frompcp-applied.Rmdusingknitr::rmarkdown

Last update: 2025-04-02
Started: 2025-03-20

Quickstart: applying PCP to a simulated environmental mixture
Simulating data | Model selection | Grid search for parameter tuning | Running PCP | Benchmarking with PCA | Evaluating PCP against the ground truth | After PCP
Rendered frompcp-quickstart.Rmdusingknitr::rmarkdown

Last update: 2025-04-02
Started: 2025-03-20

Theory crash course
PCP modeling overview | The low-rank matrix | The sparse matrix | Extensions for environmental health data | Missing value functionality | Leveraging potential limit of detection (LOD) information | Non-negativity constraint on the estimated L matrix | Convex vs. non-convex PCP | Convex PCP | Non-convex PCP | PCP parameters | Intuition behind lambda, mu, and eta | Theoretically optimal parameters | Empirically optimal parameters | Tuning parameters with grid_search_cv() | Cross-validation procedure | Best practices for $\xi$ and $K$ | Interpretation of results | Coded example analyses | References
Rendered fromtheory-crash-course.Rmdusingknitr::rmarkdown

Last update: 2025-03-25
Started: 2025-03-23