Spatial animations in R with ggplot2 and the animation package

One of the simplest ways to make an animation in R is to use the animation package to generate an html/javascript wrapper around a sequence of still images. In this post I’m going to demonstrate how to create this kind of animation for spatial data using ggplot2 and ggmap to plot point data on a Google Maps background.

Here I’ve used crime data extracted from data.police.uk, consisting of date, locations and categories of crimes commited in the Liverpool area during 2016.

First, load the packages and data. The data is in the format of one row per crime, and I’ve filtered it to include only the columns of interest for this exercise.

library(ggplot2)
library(ggmap)
library(animation)

# Read the data: Crimes in the Liverpool area during 2016 
# (Contains public sector information licensed under the Open Government Licence v3.0)
# Data for Merseyside Police, January to December 2016 obtained from https://data.police.uk/
# Grab my edit for Liverpool area with extraneous columns removed:
df=read.csv("https://github.com/annezj/basic_R_tutorials/raw/master/data/Liverpool-01-2016-12-2016.csv")

Next, get a Google Maps background for the region of interest.

latmin=min(df$Latitude)
latmax=max(df$Latitude)
lonmin=min(df$Longitude)
lonmax=max(df$Longitude)
mymap<-get_map(location=c(lonmin,latmin,lonmax,latmax)) 

The animation will be one frame per month, so process the date string to get month numbers and list month names ready for labelling.

df$monthnum=as.integer(substr(df$Month, 6, 7))
monthstrings=c("January", "February", "March", "April", "May", 					"June", "July", "August", "September",
               "October", "November", "December")
               df$monthname=factor(df$monthnum,levels=1:12,labels=monthstrings)

A quick plot confirms the data has loaded correctly and indicates the approximate number of crimes we’ll expect to see per category and frame. We also see there’s some seasonal variation which is different for different categories.

ggplot(df)+
  geom_bar(aes(x=monthname, fill=Crime.type), position="stack")+
  theme_bw()+ylab("Number of crimes")+
  xlab("")

Now on to the animation, which will have 12 frames: one per month. Define the plotting function required by the animation package. This function must create one plot per frame.

create.plots <-function()
{
  # Set an frame rate of 1 per second
  ani.options(interval=1, verbose=FALSE)
  # animvec is used to loop over frames
  animvec=1:nframes

  # loop over months/frames
  for(iframe in animvec)
  {
    # Pick up crimes occurring this month
    dfi=df[which(df$monthnum==iframe),]
    
    # Print the base map and title, implement various appearance tweaks
    titlestr=paste(monthstrings[iframe],'\n',2016,sep='')
    p=ggmap(mymap)+ggtitle(titlestr)+theme_bw()+
      theme(plot.title = element_text(hjust = 0.5),text = element_text(size=16))+
  	theme(axis.title.x=element_blank(),axis.text.x=element_blank(),axis.ticks.x=element_blank())+
      theme(axis.title.y=element_blank(),axis.text.y=element_blank(),axis.ticks.y=element_blank())
    
    # Plot all crimes for this frame
    p=p+geom_point(data=dfi,aes(x=Longitude,y=Latitude,color=Crime.type), alpha=0.4)
    
    # Fix the color scale across all frames so no categories are dropped
    p=p+scale_color_discrete("", drop=F)
    
    # Fix transparency, size of points and spacing on the legend
    p=p+guides(colour = guide_legend(override.aes = list(alpha = 1, size=4), keyheight = 1.5))
    print(p)  
  }
}

Animate by calling the function to create an html animation from the plotting function. (Note that the animation package has functions with very similar syntax to create other animation formats such as .gif, .mp4, but these require additional tools to support the conversion.)

saveHTML(create.plots(),
         img.name="crime_animation",
         autoplay=T,
         outdir=getwd(),
         htmlfile=paste("crime_animation.html", sep=""),
         ani.height=800,ani.width=800,
         title="Crime Animation",
         description=c("none")
)

As configured here the html file will appear in the current working directory and open automatically in the default web browser. The html, javascript and css directory structure can then be copied for sharing.

The finished animation looks like this.

It’s not the prettiest animation wrapper, but good enough for a quick demo that can easily be run locally or shared on the web. Is this the best way to visualise the data? Probably not. It would be much more interesting to aggregate the points by post code or on a raster grid and create a risk map, and of course it would be nice to make the map interactive. I’ll return to discuss these in a future post.