Scripts/test.R

In MarcSchneble/geonet: Intensity Estimation on Geometric Networks with Penalized Splines

document()
load_all()
library(spatstat)

X <- runifgn(50, small_gn)
delta <- "0"
h <- 0.2
r <- 2
formula <- ~1
model <- intensity_pspline(X, delta = delta, h = h, r = r)
Y <- rgnpp(50, model)
model2 <- intensity_kernel(X)

X <- runifgn(50, small_gn)
delta <- 0.2
h <- 0.1
r <- 1
model <- intensity_pspline(X, delta = delta, h = h, r = 2, verbose = TRUE,
                           control = list(rho_start = 20,
                                          eps_theta = 1e-4,
                                          rho_max = 100))
summary(model)
plot(model)

X <- as_gnpp(chicago)
model <- intensity_pspline(X, delta = 10, h = 2)
model <- intensity_kernel(X)
plot(model)
summary(model)




X <- as_gnpp(chicago)
delta <- 10
h <- 2
r <- 2
formula <- X ~ marks + x + y + dist2V

start <- Sys.time()
model <- intensity_pspline(X, formula = formula, delta = delta, h = h, r = r,
                           verbose = TRUE)
print(Sys.time() - start)

summary(model)
g <- plot(model, data = TRUE)

pdf(file = "intens_chicago_covariates.pdf", width = 6, height = 5)
print(g)
dev.off()

model <- intensity_pspline(X, delta = delta, h = h, r = r,
                           verbose = TRUE)
g <- plot(model, data = TRUE)

pdf(file = "intens_chicago.pdf", width = 6, height = 5)
print(g)
dev.off()


X <- montgomery
delta <- 0.1
h <- 0.05
r <- 2
formula <- ~ type + direction + s(hour)
formula <- ~1
model <- intensity_pspline(X, formula = formula, delta = delta, h = h, r = r, verbose = TRUE)
summary(model)
plot(model)
g2 <- plot(model)[[2]] +
  labs(x = "Hour of the day", y = "Effect on the log-scale")

pdf(file = "montgomery_hour.pdf", width = 6, height = 4)
print(g2)
dev.off()


X <- runifgn(20, small_gn)
fit1 <- intensity_pspline(X, delta = 0.2, h = 0.1)
fit2 <- intensity_pspline(X, delta = 0.5, h = 0.1)
network_intensity(fit1, fit2)

X <- as_gnpp(chicago)
fit1 <- intensity_pspline(X, verbose = TRUE)
X2 <- rgnpp(1000, fit)
fit2 <- intensity_pspline(X2, delta = 10, h = 2)
network_ISE(fit1, fit2)


X <- as_gnpp(chicago)
fit <- intensity_kernel(X)
g <- plot(fit)