Node.JS script to aggregate NYC vehicle collision data into a hexagonal grid and compute the ck means for each cell using Turf.JS and Simple Statistics.JS

.gitignore

node_modules/
.DS_Store

crashes-data-processing.js

// load libraries
const fs = require("fs")
const turf = require("turf")
const turfMeta = require("@turf/meta")
const simpleStats = require("simple-statistics")
const d3 = require("d3-array")

// load our data
const data = require("./nyc_crashes.json")

// rough geographic bounding box of NYC
const bbox = [ -74.24939, 40.50394, -73.701195, 40.90995 ]
// size in kilometers we want each side of our hex grids
const cellSize = 0.5

// create the hexbin geometry for the given bbox and cell resolution
const hexGrid = turf.hexGrid(bbox, cellSize)

// perform a "spatial join" on our hexGrid geometry and our crashes point data
const collected = turf.collect(hexGrid, data, "cartodb_id", "values")

// get rid of polygons with no joined data, to reduce our final output file size
collected.features = collected.features.filter(d => d.properties.values.length)

// count the number of crashes per hexbin
turfMeta.propEach(collected, props => {
  props.count = props.values.reduce((acc, cur) => acc += 1, 0)
})

// reduce our count values to a new array of numbers
const reduced = turfMeta.featureReduce(collected, (acc, cur) => {
  acc.push(cur.properties.count)
  return acc
}, [])

// compute the ckMeans binning for data into 7 classes from reduced values
const ck = simpleStats.ckmeans(reduced, 7)

// tack on the bin number to our data, as well as its min and max values
turfMeta.propEach(collected, props => {
  ck.forEach((bin, index) => {
    if (bin.indexOf(props.count) > -1) {
      props.bin = index
      props.binVal = d3.extent(bin)
    }
  })
})

// remove the "values" property from our hexBins as it's no longer needed
turfMeta.propEach(collected, props => {
  delete props.values
})

// write output data
fs.writeFileSync("./processed.json", JSON.stringify(collected))

nyc_crashes.json

package-lock.json

package.json

processed.json