The goal of grmbayes is to provide functions for fitting efficient
Bayesian geostatistical regression models.
You can install the development version of ensembleDownscaleR like so:
devtools::install_github("WyattGMadden/grmbayes", build_vignettes = TRUE)library(grmbayes)
?cmaq_aqs_matched
cmaq_fit <- grm(Y = cmaq_aqs_matched$pm25,
X = cmaq_aqs_matched$ctm,
L = cmaq_aqs_matched[, c("elevation", "forestcover",
"hwy_length", "lim_hwy_length",
"local_rd_length", "point_emi_any")],
M = cmaq_aqs_matched[, c("tmp", "wind")],
coords = cmaq_aqs_matched[, c("x", "y")],
n.iter = 100,
burn = 20,
thin = 4,
nngp = T,
num_neighbors = 10,
covariance = "matern",
matern.nu = 0.5,
space.id = cmaq_aqs_matched$space_id,
time.id = cmaq_aqs_matched$time_id,
spacetime.id = cmaq_aqs_matched$spacetime_id,
verbose.iter = 10)
cv_id_ctm_ord <- create_cv(space.id = cmaq_aqs_matched$space_id,
time.id = cmaq_aqs_matched$time_id,
type = "ordinary")
cmaq_fit_cv <- grm_cv(Y = cmaq_aqs_matched$pm25,
X = cmaq_aqs_matched$ctm,
cv.object = cv_id_ctm_ord,
L = cmaq_aqs_matched[, c("elevation", "forestcover",
"hwy_length", "lim_hwy_length",
"local_rd_length", "point_emi_any")],
M = cmaq_aqs_matched[, c("tmp", "wind")],
n.iter = 1000,
burn = 200,
thin = 4,
coords = cmaq_aqs_matched[, c("x", "y")],
nngp = T,
num_neighbors = 10,
covariance = "matern",
matern.nu = 0.5,
space.id = cmaq_aqs_matched$space_id,
time.id = cmaq_aqs_matched$time_id,
spacetime.id = cmaq_aqs_matched$spacetime_id,
verbose.iter = 10)
?modis_aqs_matched
cv_id_modis_ord <- create_cv(space.id = modis_aqs_matched$space_id,
time.id = modis_aqs_matched$time_id,
type = "ordinary")
modis_fit <- grm(Y = modis_aqs_matched$pm25,
X = modis_aqs_matched$aod,
L = modis_aqs_matched[, c("elevation", "forestcover",
"hwy_length", "lim_hwy_length",
"local_rd_length", "point_emi_any")],
M = modis_aqs_matched[, c("tmp", "wind", "cmaq", "tempaod",
"windaod", "elevationaod")],
n.iter = 500,
num_neighbors = 6,
burn = 100,
thin = 4,
coords = modis_aqs_matched[, c("x", "y")],
space.id = modis_aqs_matched$space_id,
time.id = modis_aqs_matched$time_id,
spacetime.id = modis_aqs_matched$spacetime_id)
modis_fit_cv <- grm_cv(Y = modis_aqs_matched$pm25,
X = modis_aqs_matched$aod,
cv.object = cv_id_maia_ord,
L = modis_aqs_matched[, c("elevation", "forestcover",
"hwy_length", "lim_hwy_length",
"local_rd_length", "point_emi_any")],
M = modis_aqs_matched[, c("tmp", "wind", "cmaq", "tempaod",
"windaod", "elevationaod")],
n.iter = 500,
burn = 100,
thin = 4,
coords = modis_aqs_matched[, c("x", "y")],
space.id = modis_aqs_matched$space_id,
time.id = modis_aqs_matched$time_id,
spacetime.id = modis_aqs_matched$spacetime_id)?cmaq_full
cmaq_pred <- grm_pred(grm.fit = cmaq_fit,
X.pred = cmaq_full$ctm,
L.pred = cmaq_full[, c("elevation", "forestcover",
"hwy_length", "lim_hwy_length",
"local_rd_length", "point_emi_any")],
M.pred = cmaq_full[, c("tmp", "wind")],
coords.Y = cmaq_aqs_matched[, c("x", "y")],
space.id.Y = cmaq_aqs_matched$space_id,
coords.pred = cmaq_full[, c("x", "y")],
space.id = cmaq_full$space_id,
time.id = cmaq_full$time_id,
spacetime.id = cmaq_full$spacetime_id,
n.iter = 20,
verbose = T,
include.random.effects = T)
?modis_full
modis_pred <- grm_pred(grm.fit = modis_fit,
X.pred = modis_full$aod,
L.pred = modis_full[, c("elevation", "forestcover",
"hwy_length", "lim_hwy_length",
"local_rd_length", "point_emi_any")],
M.pred = modis_full[, c("tmp", "wind", "cmaq", "tempaod",
"windaod", "elevationaod")],
coords.Y = modis_aqs_matched[, c("x", "y")],
space.id.Y = modis_aqs_matched$space_id,
coords.pred = modis_full[, c("x", "y")],
space.id = modis_full$space_id,
time.id = modis_full$time_id,
spacetime.id = modis_full$spacetime_id,
n.iter = 100,
verbose = T)