How to draw flows in ggplot2 based on a flow matrix

StackOverflow https://stackoverflow.com/questions/15191898

  •  18-03-2022
  •  | 
  •  

Question

I have commuter flow data from the 2001 Census. I converted it from "flat" to "long" form using melt from the reshape2 R package and placed origins and destinations in the same row because ggplot2 accepts only data-frame inputs.

My issue is that I've ended-up doubling-up the data for each line, so that each row (line) has an origin and a destination. I'm sure there is a more concise solution, probably involving an even longer form of my data.

To make the problem specific, I've produced a small worked example from Hereford:

# prepare data + packages
library(ggplot2)
library(ggmap)
flows.mini <- flows.ft[1:100,]
save(flows.mini, file="flows.mini.RData")
load("flows.mini.RData")
head(flows.mini)

 variable    X.1 value      X1.x     X2.x      X1.y     X2.y   n nr
1   00GANY 00GANY   605 -2.699389 52.06554 -2.699389 52.06554 605  1
2   00GANY 00GAPA   135 -2.742064 52.04099 -2.699389 52.06554 135  2
3   00GANY 00GAQD    25 -2.733890 51.93402 -2.699389 52.06554  25  3
    fcols
1    500+
2 100-500
3  10-100

To reproduce the steps taken from the last two lines, please download the RData file (2 kb): http://dl.dropbox.com/u/15008199/flows.mini.RData and reproduce the plot:

mini-flows-plot

This is how I've plotted it:

# plot flows by doubling-up
hford <- qmap("hereford", source = "stamen", maptype = "toner", extent = "normal", maprange=FALSE)

hford + geom_path(data= flows.mini, aes(x=c(X1.x,X1.y), y=c(X2.x, X2.y), 
      group = c(nr, nr), color = c(fcols,fcols), size= c(n,n)), 
      lineend = "round") +
  scale_size_continuous(range = c(0.05,5)) +
  scale_color_discrete(breaks = c("0-10", "10-100", "100-500", "500+")) + 
  coord_map()

I think you'll agree the double attributes are inefficient, so, to re-phrase my question: how can I remove them?

Was it helpful?

Solution

I'm not sure I really understand the question, but if you use geom_segment instead of geom_path, does it solve the problem you describe of "doubling the attributes" ?

Something like :

hford + geom_segment(data= flows.mini, aes(x=X1.x,y=X2.x, xend=X1.y, yend=X2.y, 
      group = nr, color = fcols, size= n), 
      lineend = "round") +
  scale_size_continuous(range = c(0.05,5)) +
  scale_color_discrete(breaks = c("0-10", "10-100", "100-500", "500+")) + 
  coord_map()

enter image description here

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top