A minimal map of Vancouver’s street trees

Ideas


I have been a long-time admirer of the ‘Minimal Maps’ series from Michael Pecirno, and wanted to try to recreate something similar. I really liked his focus on the natural environment and spent a while exploring potential datasets, at a variety of scales. I eventually settled on this dataset of street trees in Vancouver. This dataset has one point for every public tree on Vancouver’s boulevards (there are over 140,000!).

Data Cleaning

I downloaded this dataset as a CSV (it was taking too long to open in R as a GeoJSON) and read it into R.

# loading in necessary libraries 
library(ggplot2)
library(rlist)
library(sf)

# open the dataset 
trees <- read.csv('street-trees.csv', sep = ';')

Because I am still finding R’s factor data type challenging to work with, I converted everything into integers or strings. I also removed the entries without any coordinates.

# convert factors to strings 
i <- sapply(trees, is.factor) # find columns that are factors 
trees[i] <- lapply(trees[i], as.character) # convert column to character

# remove rows without coordinates 
new <- subset(trees, nchar(Geom) > 1)

The most challenging part of this process was getting the coordinate data into a workable format. Initially, this data for each tree was stored in a cell in a format like this:

{\“type\”: \“Point\“, \“coordinates\”: [-123.135032, 49.230858]}

I wrote a function to parse these strings, one at a time, and return a list that contained just the coordinates.

# function to parse strings to get coordinates
getcoords <- function(string){
  
  split <- strsplit(string,']')[[1]]
  a <- split[[1]]
  b <- strsplit(a, '-')[[1]][[2]]
  c <- strsplit(b, ',')
  d <- c[[1]]
  
  x <- paste0('-', d[[1]])
  y <- strsplit(d[[2]], ' ')[[1]][[2]]
  
  coords <- list(x,y)
  return(coords)
  
}

I applied this function to each point in the dataset and created new columns for each coordinate.

# apply the function for each row 
for (i in 1:length(new$Geom)){
  
  # parse to get the coordinates 
  z <- getcoords(new$Geom[[i]])
  x <- z[[1]]
  y <- z[[2]]
  
  # add new columns to df 
  new$lats[i] <- x
  new$longs[i] <- y

}

# convert the columns from characters to nums 
new$lats <- as.numeric(new$lats)
new$longs<- as.numeric(new$longs)

Visualizing the data

Once all the data was cleaned, it was a very simple matter to visualize using ggplot2. Given that the dataset was decently large, I took a smaller subset of all points for faster visualization while I tested different visualization parameters.

# get smaller version for faster testing 
small <- new[1:2500,]

# make the map 
map <- ggplot(data=small) +
  geom_point(data=small, 
             aes(x=small$lats, 
                 y=small$longs), 
             size=small$DIAMETER*0.1, 
             alpha=0.005, 
             color='#ffffff')+ 
  theme_map()

After outputting this map, I used inDesign to add some simple text for a title.