spclustr

Rendered reports: mape_r-01 · mape_r-01_reto

spclustr is a personal learning lab of standalone R scripts reproducing spatial statistics tutorials, package vignettes, and papers. It is not an R package — each .R/.Rmd file is a self-contained reproduction covering one method or one package, spanning point pattern analysis, areal/disease mapping, geostatistics, GAM-based smoothing, and scan statistics for cluster detection (e.g. via spatstat, spdep, rsatscan).

What’s inside

The scripts are grouped below by the spatial-statistics theme they demonstrate. Each entry lists the script(s) and the key package(s) exercised.

Point pattern analysis

Intensity/density estimation, case-control (bivariate) patterns, spatio-temporal point processes, relative risk mapping.

• dc-spatstat-c2.R, dc-spatstat-c2.5.R — DataCamp “Spatial Statistics in R”, ch. 2: bivariate/case-control point patterns, spatio-temporal descriptive & inferential analysis, ppp ↔ sf conversion. (spatstat, sf, tmap, raster, spatialkernel, splancs, ggspatial, maptools)
• mape_r-01.R / mape_r-01.Rmd — teaching session on bivariate point pattern density & relative-risk mapping, prepared for Club de Análisis Espacial, Sesión #5 (“mapear”). (spatstat, sf, maptools, tidyverse)
  • Rendered: html
• mape_r-01_reto.Rmd — challenge notebook for the same session: mapping geolocated COVID-19 China case data as sf. (sf, ggspatial, googlesheets4, janitor, tidyverse)
  • Rendered: html

Areal data & disease mapping

Choropleths, cartograms, Moran’s I spatial autocorrelation, spatial GLM regression, Bayesian (BYM/INLA) disease mapping.

• dc-spatstat-c3.R — DataCamp ch. 3, London EU referendum data: choropleths, cartograms, Moran’s I, SIR/GLM regression. (sf, tmap, raster, cartogram, rgeos, spdep, R2BayesX)
• moraga-RJ-2018-036.R — reproduction of Paula Moraga (2018), Small Area Disease Risk Estimation and Visualization Using R: SIR calculation, neighborhood matrix, Bayesian disease mapping with INLA (BYM model). (SpatialEpi, spdep, INLA, sf)

Find a recent implementation of Moran’s Index spatial autocorrelation, as well as High-High and Low-Low clusters (for spatial correlation and incidence rate), for spatiotemporal analysis with {epiCo}

Geostatistics

Trend surfaces, variograms, kriging.

• dc-spatstat-c4.R — DataCamp ch. 4: trend surface analysis, variograms, kriging (manual and via automap). (spatstat, gstat, raster, automap, broom)

GAM-based spatial smoothing

Generalized additive models for smoothing spatial risk/odds surfaces.

• dc-spatstat-gam_reprex.R — case-control intensity/relative-risk mapping followed by GAM modeling of spatial risk. (spatstat, mgcv, mgcViz, geoR, fields, viridis)
• gam-mapgam-reprex.R — mapGAM/modgam smoothing of odds ratios & relative risk over a spatial prediction grid. (mapGAM, PBSmapping, maps, mapproj)
• gam-reprex.R — general GAM modeling walkthrough: smooths, diagnostics. (mgcv, mgcViz, gamair, broom, MASS)

Scan statistics for cluster detection

Spatial and space-time cluster detection.

• satscan-0-vignette.R — rsatscan vignette, wrapping the external SaTScan software: Bernoulli spatial, Poisson space-time/discrete-time, and space-time permutation scan statistics. (rsatscan, spatstat, sf, janitor)
• scanstatistics-1-vignette.R — scanstatistics package vignette: expectation-based Poisson scan statistics for space-time cluster detection. (scanstatistics, sp, tsibble, magrittr)

Supporting utilities

• ggmap-0-vignette.R — ggmap vignette: basemap tiles and plotting geolocated crime points. (ggmap, tidyverse)

Installation

This repo is a flat collection of scripts, not an installable R package — there is no DESCRIPTION or renv.lock. To run a script:

# 1. clone the repo, then open spclustr.Rproj in RStudio
# 2. install the packages required by the script you want to run, e.g.:
install.packages(c("spatstat", "sf", "tmap", "spdep", "gstat", "mgcv"))

# some scripts need packages outside CRAN:
# - INLA (used by moraga-RJ-2018-036.R):
install.packages("INLA", repos = "https://inla.r-inla-download.org/R/stable")
# - rsatscan (used by satscan-0-vignette.R) also requires the standalone
#   SaTScan desktop application installed separately: https://www.satscan.org/

# 3. run the script

References

• Moraga, P. (2018). Small Area Disease Risk Estimation and Visualization Using R. The R Journal. link
• SaTScan software & rsatscan package: https://www.satscan.org/rsatscan/rsatscan.html
• scanstatistics package vignette: expectation-based scan statistics for space-time cluster detection
• mapGAM package: GAM-based smoothing for spatial epidemiology
• DataCamp, “Spatial Statistics in R” course (chapters 2–4 reproduced here)

Author

• Andree Valle Campos (@avallecam)

Acknowledgements

Some material here was originally prepared for Club de Análisis Espacial, a spatial-analysis study group, as teaching/challenge material for Sesión #5 (“mapear”).

License

MIT