In cmjt/stelfi: Hawkes and Log-Gaussian Cox Point Processes Using Template Model Builder

Datasets in stelfi {#data}

Below are the data packaged within stelfi.

library(stelfi)
data(package = "stelfi")$result[, c("Item", "Title")]

Temporal point pattern data

## load the tidyverse packages
library(tidyverse)

retweets_niwa {-}

In 2019 a NIWA scientist found a working USB in the scat of a leopard seal, they then tweeted about it in the hopes of finding its owner. In this chapter a Hawkes process is fitted to these data.

##devtools::install_github("gadenbuie/tweetrmd")
library(tweetrmd)
include_tweet("https://twitter.com/niwa_nz/status/1092610541401587712")

The retweets_niwa dataset contains the retweet timestamps for this tweet.

data(retweets_niwa, package = "stelfi")
head(retweets_niwa)

ggplot(data.frame(time = retweets_niwa), aes(x = time)) + 
  geom_histogram() + ylab("Retweet frequency") + xlab("") +
  theme_bw()

uk_serial {-}

Murder UK documents some of the UKs most infamous multiple murderer cases. The uk_serial dataset contains summary information about the documented cases along with approximate timeframes.

data("uk_serial", package = "stelfi")
head(uk_serial)

uk_serial %>%
  mutate(time =  paste(date_of_first_kill, "/01", sep='')) %>%
  mutate(time = as.Date(time, "%m/%Y/%d")) %>%
  ggplot(aes(x = time)) + 
  geom_histogram() + 
  ylab("Frequency of known first kill") + 
  xlab("") + theme_bw()

Spatiotemporal point pattern data

Using maps to create sf objects of country boundaries:

us <- maps::map("usa", fill = TRUE, plot = FALSE) %>%
    sf::st_as_sf() %>%
    sf::st_make_valid()
nz <- maps::map("nz", fill = TRUE, plot = FALSE) %>%
    sf::st_as_sf() %>%
    sf::st_make_valid()
iraq <- maps::map("world", "Iraq", fill = TRUE, plot = FALSE) %>%
     sf::st_as_sf() %>%
    sf::st_make_valid()

sasquatch {-}

The Bigfoot Field Researchers Organization (BFRO) documents Bigfoot (Sasquatch) sightings; some data have been collated and packaged in stelfi assasquatch.

data("sasquatch", package = "stelfi")
sasquatch

## needed for GDAL shipped with older Ubuntu dist
sf::st_crs(sasquatch) <- "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0" 

ggplot(sasquatch, aes(x = date)) + geom_histogram(bins = 150) +
  ylab("Frequency of Sasquatch sightings") + xlab("") +
  theme_bw()

ggplot(sasquatch) +  
  geom_sf(alpha = 0.3) + 
  coord_sf() + 
  geom_sf(data = us, fill = NA) + 
  theme_classic()

nz_earthquakes {-}

GeoNet Quake Search catalogues New Zealand earthquake occurrence; some of these data have been and packaged in stelfi as nz_earthquakes. In this chapter a Hawkes process is fitted to these data.

data("nz_earthquakes", package = "stelfi")
nz_earthquakes

## needed for GDAL shipped with older Ubuntu dist
sf::st_crs(nz_earthquakes) <- "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0" 

ggplot(nz_earthquakes, aes(x = origintime)) + geom_histogram(bins = 100) +
  ylab("Frequency of earthquakes") + xlab("") +
  theme_bw()

ggplot(nz_earthquakes) +  
  geom_sf(alpha = 0.1) + 
  coord_sf() + 
  geom_sf(data = nz, fill = NA) +
  theme_classic()

nz_murders {-}

The Homicide Report documents homicides in New Zealand. The nz_murders dataset contains summary information about the documented cases. In this chapter a spatiotemporal self-exciting model is fitted to these data.

data("nz_murders", package = "stelfi")
nz_murders

## needed for GDAL shipped with older Ubuntu dist
sf::st_crs(nz_murders) <- "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0" 

ggplot(nz_murders, aes(x = full_date)) + geom_histogram(bins = 100) +
  ylab("Frequency of murders") + xlab("") +
  theme_bw()

iraq_terrorism {-}

The Global Terrorism Database (GTD) documents information on terrorism events worldwide; some of these data have been and packaged in stelfi as iraq_terrorism.

data("iraq_terrorism", package = "stelfi")
iraq_terrorism

## needed for GDAL shipped with older Ubuntu dist
sf::st_crs(iraq_terrorism) <- "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0" 

iraq_terrorism %>%
  mutate(date = paste(iday, imonth, iyear, sep = "/")) %>%
  mutate(date = as.Date(date, "%d/%m/%Y")) %>%
  ggplot(., aes(x = date)) + geom_histogram(bins = 150) +
  ylab("Frequency of attacks") + xlab("") +
  theme_bw()

ggplot(iraq_terrorism) +  
  geom_sf(alpha = 0.3) + 
  coord_sf() + 
  geom_sf(data = iraq, fill = NA) +
  theme_classic()

Simulated data

xyt {-}

In this chapter a log-Gaussian Cox process is fitted to these data and in this chapter a spatiotemporal selfexciting model is fitted.

data("xyt", package = "stelfi")
xyt_sf <- sf::st_as_sf(xyt)
xyt_sf

ggplot(xyt_sf) + geom_sf(fill = NA) +
  theme_void()

marked {-}

In this chapter a marked log-Gaussian Cox process is fitted to these data.

data(marked, package = "stelfi")
marked_sf <- sf::st_as_sf(x = marked,
                        coords = c("x", "y"))
marked_sf

domain <- list(3 * cbind(c(0, 1, 1, 0, 0), c(0, 0, 1, 1, 0))) %>%
  sf::st_polygon() %>% sf::st_sfc() %>% sf::st_sf(geometry = .)

ggplot(marked_sf, aes(col = m1)) + 
  geom_sf() + labs(color = "Mark") +
  scale_color_continuous(type = "viridis") +
  geom_sf(data = domain, fill = NA, inherit.aes = FALSE) +
  theme_void()

horse_mesh {-}

In this chapter we illustrate different geometric metrics of this triangulation.

data("horse_mesh", package = "stelfi")
horse_mesh_sf <- stelfi::mesh_2_sf(horse_mesh)
horse_mesh_sf

ggplot(horse_mesh_sf) + geom_sf(fill = NA, col = "black") +
  theme_void()

cmjt/stelfi documentation built on Oct. 25, 2023, 2:53 p.m.