This example shows how to use R and QGIS from within R to perform a series of common point pattern analysis techniques.
library(mapview)
library(sf)
## Linking to GEOS 3.8.0, GDAL 3.0.4, PROJ 6.3.1
library(censusxy)
library(dplyr)
##
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
##
## filter, lag
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union
addr<-read.csv(url("https://raw.githubusercontent.com/coreysparks/data/master/wic_west_side.csv"))
addr<-addr[c(6, 12:14)]
names(addr)<-c("street", "city", "st", "zip")
head(addr)
results<-cxy_geocode(addr,
street = "street",
city = "city",
state ="st",
zip = "zip",
class="sf",
output = "simple")
## 36 rows removed to create an sf object. These were addresses that the geocoder could not match.
results.proj<-st_transform(results,
crs = 2278)
addr<-read.csv(url("https://raw.githubusercontent.com/coreysparks/data/master/wic_west_side.csv"))
results <- st_as_sf(addr, coords=c("Longitude", "Latitude"), crs=4269,agr="constant")
results.proj<-st_transform(results,
crs = 2278)
mapview(results.proj)
mean_feature<-apply(st_coordinates(results.proj), MARGIN = 2, FUN = mean)
mean_feature<-data.frame(place="meanfeature", x=mean_feature[1], y= mean_feature[2])
mean_feature<-st_as_sf(mean_feature, coords = c("x", "y"), crs= 2278)
mapview(mean_feature, col.regions="red")+mapview( results)
median_feature<-apply(st_coordinates(results.proj), MARGIN = 2, FUN = median)
median_feature<-data.frame(place="medianfeature", x=median_feature[1], y= median_feature[2])
median_feature<-st_as_sf(median_feature, coords = c("x", "y"), crs= 2278)
mapview(median_feature, col.regions="green")+
mapview(mean_feature, col.regions="red")+
mapview( results)
wicbuff<- st_buffer(results.proj, dist = 2500)
mapview(wicbuff)+mapview(results.proj, col.regions="green")
chull <- st_convex_hull(st_union(results))
## although coordinates are longitude/latitude, st_union assumes that they are planar
mapview(chull)+
mapview(results, col.regions = "green")
R can do kernel density maps, but using simple features it’s kind of complicated. I will use Qgis through R instead using the qgisprocess package
library(qgisprocess)
## Using 'qgis_process' at 'C://OSGeo4W64/bin/qgis_process-qgis.bat'.
## Run `qgis_configure()` for details.
qgis_configure()
## getOption('qgisprocess.path') was not found.
## Sys.getenv('R_QGISPROCESS_PATH') was not found.
## Trying 'qgis_process' on PATH
## Error in rethrow_call(c_processx_exec, command, c(command, args), stdin, : Command 'qgis_process' not found @win/processx.c:994 (processx_exec)
## Found 2 QGIS installations containing 'qgis_process':
## C://OSGeo4W64/bin/qgis_process-qgis.bat
## C://OSGeo4W64/bin/qgis_process-qgis-dev.bat
## Trying command 'C://OSGeo4W64/bin/qgis_process-qgis.bat'
## Success!
To use this, we need to find the name of the Qgis algorithm we want. qgis_algorithms() can return all available algorithms, then we can either filter it with View() or use grep to search for one.
algs<-qgis_algorithms()
algs[grepl(pattern = "density", x = algs$algorithm ),]
qgis_show_help("qgis:heatmapkerneldensityestimation")
## C:\Users\ozd504\Documents\GitHub\DEM7093\code
## Heatmap (Kernel Density Estimation) (qgis:heatmapkerneldensityestimation)
##
## ----------------
## Description
## ----------------
##
## ----------------
## Arguments
## ----------------
##
## INPUT: Point layer
## Argument type: source
## Acceptable values:
## - Path to a vector layer
## RADIUS: Radius
## Argument type: distance
## Acceptable values:
## - A numeric value
## RADIUS_FIELD: Radius from field
## Argument type: field
## Acceptable values:
## - The name of an existing field
## - ; delimited list of existing field names
## PIXEL_SIZE: Output raster size
## Argument type: number
## Acceptable values:
## - A numeric value
## WEIGHT_FIELD: Weight from field
## Argument type: field
## Acceptable values:
## - The name of an existing field
## - ; delimited list of existing field names
## KERNEL: Kernel shape
## Argument type: enum
## Available values:
## - 0: Quartic
## - 1: Triangular
## - 2: Uniform
## - 3: Triweight
## - 4: Epanechnikov
## Acceptable values:
## - Number of selected option, e.g. '1'
## - Comma separated list of options, e.g. '1,3'
## DECAY: Decay ratio (Triangular kernels only)
## Argument type: number
## Acceptable values:
## - A numeric value
## OUTPUT_VALUE: Output value scaling
## Argument type: enum
## Available values:
## - 0: Raw
## - 1: Scaled
## Acceptable values:
## - Number of selected option, e.g. '1'
## - Comma separated list of options, e.g. '1,3'
## OUTPUT: Heatmap
## Argument type: rasterDestination
## Acceptable values:
## - Path for new raster layer
##
## ----------------
## Outputs
## ----------------
##
## OUTPUT: <outputRaster>
## Heatmap
Run the algorithm
wic_dens<-qgis_run_algorithm(algorithm ="qgis:heatmapkerneldensityestimation",
INPUT=results.proj,
RADIUS = 5000,
PIXEL_SIZE = 100,
KERNEL = 0,
OUTPUT=file.path(getwd(), "wicdenst.TIF"),
load_output = TRUE)
## Running "C://OSGeo4W64/bin/qgis_process-qgis.bat" run \
## "qgis:heatmapkerneldensityestimation" \
## "--INPUT=C:\Users\ozd504\AppData\Local\Temp\RtmpmoOrgu\file202815a448f1\file202843437b8.gpkg" \
## "--RADIUS=5000" "--PIXEL_SIZE=100" "--KERNEL=0" "--OUTPUT_VALUE=0" \
## "--OUTPUT=C:/Users/ozd504/Documents/GitHub/DEM7093/code/wicdenst.TIF"
## C:\Users\ozd504\Documents\GitHub\DEM7093\code
##
## ----------------
## Inputs
## ----------------
##
## INPUT: C:\Users\ozd504\AppData\Local\Temp\RtmpmoOrgu\file202815a448f1\file202843437b8.gpkg
## KERNEL: 0
## OUTPUT: C:/Users/ozd504/Documents/GitHub/DEM7093/code/wicdenst.TIF
## OUTPUT_VALUE: 0
## PIXEL_SIZE: 100
## RADIUS: 5000
##
##
## 0...10...20...30...40...50...60...70...80...90...
## ----------------
## Results
## ----------------
##
## OUTPUT: C:/Users/ozd504/Documents/GitHub/DEM7093/code/wicdenst.TIF
library(raster)
library(RColorBrewer)
result<- qgis_as_raster(wic_dens)
projection(result)<-crs(results.proj)
mapview(result)+mapview(results.proj)
A spatial join can combine attributes of one layer with another layer. Here I combine census variables with the WIC clinic points.
library(tidycensus)
library(dplyr)
#load census tract data
sa_acs<-get_acs(geography = "tract",
state="TX",
county = "Bexar",
year = 2019,
variables=c( "DP05_0001E", "DP03_0009P", "DP03_0062E", "DP03_0119PE",
"DP05_0001E","DP02_0009PE","DP02_0008PE","DP02_0040E","DP02_0038E",
"DP02_0066PE","DP02_0067PE","DP02_0080PE","DP02_0092PE",
"DP03_0005PE","DP03_0028PE","DP03_0062E","DP03_0099PE","DP03_0101PE",
"DP03_0119PE","DP04_0046PE","DP05_0072PE","DP05_0073PE",
"DP05_0066PE", "DP05_0072PE", "DP02_0113PE") ,
geometry = T, output = "wide")
## Getting data from the 2015-2019 5-year ACS
## Downloading feature geometry from the Census website. To cache shapefiles for use in future sessions, set `options(tigris_use_cache = TRUE)`.
## Using the ACS Data Profile
## Using the ACS Data Profile
#rename variables and filter missing cases
sa_acs2<-sa_acs%>%
mutate(totpop= DP05_0001E, pwhite=DP05_0072PE,
pblack=DP05_0073PE , phisp=DP05_0066PE,
phsormore=DP02_0066PE,punemp=DP03_0009PE, medhhinc=DP03_0062E,
ppov=DP03_0119PE)%>%
dplyr::select(GEOID, totpop, pblack, pwhite, phisp, punemp, medhhinc, ppov)
sa_acs2<-st_transform(sa_acs2, crs = 2278)
sa_trol<-st_cast(sa_acs2, "MULTILINESTRING")
spjoin<-st_join(results.proj, sa_acs2)
head(spjoin)
mapview(spjoin["punemp"])+mapview(sa_trol)
Point in polygon operations are actually a spatial intersection (more on this next week!) where we see how many points fall within a given polygon.
sa_acs2$nwic<- lengths(st_intersects(sa_acs2, results.proj))
mapview(sa_acs2, zcol="nwic")+
mapview(results.proj, col.regions = "green")
Thiessen or Voronoi polygons are a process where we can convert points into polygons.
algs[grepl(pattern = "voronoi", x = algs$algorithm ),]
qgis_show_help("qgis:voronoipolygons")
## C:\Users\ozd504\Documents\GitHub\DEM7093\code
## Voronoi polygons (qgis:voronoipolygons)
##
## ----------------
## Description
## ----------------
## This algorithm takes a points layer and generates a polygon layer containing the voronoi polygons corresponding to those input points.
##
##
## ----------------
## Arguments
## ----------------
##
## INPUT: Input layer
## Argument type: source
## Acceptable values:
## - Path to a vector layer
## BUFFER: Buffer region (% of extent)
## Argument type: number
## Acceptable values:
## - A numeric value
## OUTPUT: Voronoi polygons
## Argument type: sink
## Acceptable values:
## - Path for new vector layer
##
## ----------------
## Outputs
## ----------------
##
## OUTPUT: <outputVector>
## Voronoi polygons
wic_von<-qgis_run_algorithm(alg="qgis:voronoipolygons",
INPUT=results.proj,
OUTPUT=file.path(tempdir(), "wicvon.shp"),
load_output = TRUE)
wic_von<-sf::read_sf(qgis_output(wic_von, "OUTPUT"))
mapview(wic_von, alpha=.75)+
mapview(results.proj, col.regions="green")
library(spatstat)
wic.pp<-as.ppp(as(results.proj, "Spatial"))
plot(nearest.neighbour(wic.pp))
algs[grepl(pattern = "nearest", x = algs$algorithm ),]
qgis_show_help("native:nearestneighbouranalysis")
## C:\Users\ozd504\Documents\GitHub\DEM7093\code
## Nearest neighbour analysis (native:nearestneighbouranalysis)
##
## ----------------
## Description
## ----------------
## This algorithm performs nearest neighbor analysis for a point layer.
##
## The output describes how the data are distributed (clustered, randomly or distributed).
##
## Output is generated as an HTML file with the computed statistical values.
##
## ----------------
## Arguments
## ----------------
##
## INPUT: Input layer
## Argument type: source
## Acceptable values:
## - Path to a vector layer
## OUTPUT_HTML_FILE: Nearest neighbour
## Argument type: fileDestination
## Acceptable values:
## - Path for new file
##
## ----------------
## Outputs
## ----------------
##
## OUTPUT_HTML_FILE: <outputHtml>
## Nearest neighbour
## OBSERVED_MD: <outputNumber>
## Observed mean distance
## EXPECTED_MD: <outputNumber>
## Expected mean distance
## NN_INDEX: <outputNumber>
## Nearest neighbour index
## POINT_COUNT: <outputNumber>
## Number of points
## Z_SCORE: <outputNumber>
## Z-score
wic_nn<-qgis_run_algorithm(alg="native:nearestneighbouranalysis",
INPUT=results.proj,
OUTPUT_HTML_FILE=file.path(tempdir(), "wicnn.html"),
load_output = TRUE)
## Ignoring unknown input 'load_output'
## Running "C://OSGeo4W64/bin/qgis_process-qgis.bat" run \
## "native:nearestneighbouranalysis" \
## "--INPUT=C:\Users\ozd504\AppData\Local\Temp\RtmpmoOrgu\file202815a448f1\file202872c11db.gpkg" \
## "--OUTPUT_HTML_FILE=C:\Users\ozd504\AppData\Local\Temp\RtmpmoOrgu/wicnn.html"
## C:\Users\ozd504\Documents\GitHub\DEM7093\code
## The system cannot find the file C:\rtools40\Version.txt.
##
## ----------------
## Inputs
## ----------------
##
## INPUT: C:\Users\ozd504\AppData\Local\Temp\RtmpmoOrgu\file202815a448f1\file202872c11db.gpkg
## OUTPUT_HTML_FILE: C:\Users\ozd504\AppData\Local\Temp\RtmpmoOrgu/wicnn.html
##
##
## 0...10...20...30...40...50...60...70...80...90...100 - done.
##
## ----------------
## Results
## ----------------
##
## EXPECTED_MD: 1356.5835389412343
## NN_INDEX: 0.4987323168352079
## OBSERVED_MD: 676.5720513566673
## OUTPUT_HTML_FILE: C:\Users\ozd504\AppData\Local\Temp\RtmpmoOrgu/wicnn.html
## POINT_COUNT: 100
## Z_SCORE: -9.589602141964955