Besides being able to use third-party tools, another main reason why I tend to use R for all my data analysis tasks is that R is extremely powerful in creating custom data exploration, visualization, and modeling designs.

As an example, let's create a flow-map based on our data, where we will highlight the flights from Houston based on the number of actual and cancelled flights. We will use lines and circles to render these two variables on a 2-dimensional map, and we will also add a contour plot in the background based on the average time delay.

But, as usual, let's do some data transformations first! To keep the number of flows at a minimal level, let's get rid of the airports outside the USA at last:

> dt <- dt[point.in.polygon(dt$lon, dt$lat,
+                           usa_data$x, usa_data$y) == 1, ]

We will need the diagram package (to render curved arrows from Houston to the destination airports) and the scales package to create transparent colors:

> library(diagram)
> library(scales)

Then, let's render the contour map described in the Contour Lines section:

> map("usa")
> title('Number of flights, cancellations and delays from Houston')
> image(look, add = TRUE)
> map("state", lwd = 3, add = TRUE)

And then add a curved line from Houston to each of the destination airports, where the width of the line represents the number of cancelled flights and the diameter of the target circles shows the number of actual flights:

> for (i in 1:nrow(dt)) {
+   curvedarrow(
+     from       = rev(as.numeric(h)),
+     to         = as.numeric(dt[i, c('lon', 'lat')]),
+     arr.pos    = 1,
+     arr.type   = 'circle',
+     curve      = 0.1,
+     arr.col    = alpha('black', dt$N[i] / max(dt$N)),
+     arr.length = dt$N[i] / max(dt$N),
+     lwd        = dt$Cancelled[i] / max(dt$Cancelled) * 25,
+     lcol       = alpha('black',
+                    dt$Cancelled[i] / max(dt$Cancelled)))
+ }

Well, this chapter ended up being about visualizing spatial data, and not really about analyzing spatial data by fitting models, filtering raw data, and looking for spatial effects. In the last section of the chapter, let's see how one can start using analytical approaches with spatial data.