In gmcirco/crimtools:

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "70%"
)

crimtools

crimtools is a collection of packages intended to make it easier for researchers to perform common data analyses on spatial data. This package interfaces with spatial data in sf format and provides functions to help plotting static maps via ggplot2 or interactive maps via maptools. This is also intended to serve as a centralized point for basic spatial analyses as many of the previous packages are no longer being updated (such as GISTools) or are being depricated in favor of sf, stars and terra (see here ). The core functionality of crimtools is based around either raster or areal analyses. While this package is still in active development, the intended main functions will be:

1. Kernel density mapping
2. Spatial grids and grid counts
3. LISA statistics (Moran's I, Getis Ord Gi*)

In the future, other functionalities provided in commercial tools (i.e. ArcGIS) will be ported to crimtools.

Installation

You can install the development version of crimtools from GitHub with:

# install.packages("devtools")
devtools::install_github("gmcirco/crimtools")

Tools Available

The crimtools package is intended to be used with simple features from the sf package (see this link for a more in-depth description of simple features in R). Currently, the following analyses are supported:

1. Kernel density estimation
2. DBCAN point clustering

Kernel Density Estimation{#kde}

The kernel_density function requires, at a minimum, a point feature x and an enclosing polygon feature region. Both of these should be sf objects and should (ideally) be projected into feet or meters. By default, kernel_density will automatically estimate a bandwidth by using the bw.ppl function as part of spatstat.core. However, users can (and should) provide a bandwidth size that reflects domain knowledge about the relevant feature. Users can also provide a custom function to estimate the bandwidth, or use several of those available in spatstat.

# Load necessary libraries
library(crimtools)
library(sf)

# Load example data
data("newhaven")
data("nh_hom")

# Default output, gaussian kernel
kde_out <- kernel_density(x = nh_hom, region = newhaven)

The kernel type, bandwidth, and other features can be manually adjusted as well.

# Specifying a quartic kernel with 2000 foot bandwidth
kde_out2 <- kernel_density(x = nh_hom, 
                           region = newhaven, 
                           kernel = "quartic", 
                           bdw = 2000)

By default kernel_density returns a dataframe with columns for the density value and the x and y coordinates of cell centroid.

head(kde_out)

The exported dataframe can easily be plotted using ggplot2 functionality, or used as a variable in an analysis. This allows for easy customization using the large number of options available in ggplot2

library(ggplot2)

# plot as a ggplot2 object using geom_tile
# and setting coord_equal
ggplot(kde_out) +
  geom_tile(aes(x = X, y = Y, fill = density)) +
  coord_equal() +
  scale_fill_viridis_c(option = "magma") +
  theme_void()

DBSCAN Point Clustering {#kde}

Given a region and point pattern, this function computes clusters based on user-defined search parameters. Here, this is the dist measure and minPts measure. This means we will search for clusters that are within a 2000 foot radius of a 'core' point, and contain a minimum of 5 points. By default, this function will output an image displaying the number of identified clusters and 'noise' points, as well as an sf point object with the cluster IDs attached.

# Run DBSCAN
cluster_out <- cluster_points(x = nh_hom, region = newhaven, dist = 2000, minPts = 5)

# check cluster ids
table(cluster_out$K)

gmcirco/crimtools documentation built on May 22, 2022, 1:31 a.m.