Dataviz at Office for National Statistics (UK)
positionedData = new AccurateBeeswarm(
data,
d => rScale(d.pop) + .2,
d => xScale(d.pct_change)
).calculateYPositions()
data = mobility_data.filter(d => d.in_lockdown === "TRUE")
class AccurateBeeswarm {
constructor(items, radiusFun, xFun) {
this.items = items;
this.radiusFun = radiusFun;
this.xFun = xFun;
this.tieBreakFn = this._sfc32(0x9E3779B9, 0x243F6A88, 0xB7E15162, seed);
this.maxR = Math.max(...items.map(d => radiusFun(d)));
this.rng = this._sfc32(1, 2, 3, seed);
}
calculateYPositions() {
let all = this.items
.map((d, i) => ({
datum: d,
originalIndex: i,
x: this.xFun(d),
y: null,
placed: false,
forbiddenY: new UnionOfIntervals(),
bestPosition: 0,
heapPos: -1
}))
.sort((a, b) => a.x - b.x);
all.forEach(function(d, i) {
d.index = i;
});
let tieBreakFn = this.tieBreakFn;
all.forEach(function(d) {
d.tieBreaker = tieBreakFn(d.x);
});
let allSortedByPriority = [...all].sort((a, b) => {
let key_a = this.radiusFun(a.datum) + a.tieBreaker * randomness1;
let key_b = this.radiusFun(b.datum) + b.tieBreaker * randomness1;
return key_b - key_a;
});
for (let item of allSortedByPriority) {
item.placed = true;
// If nothing placed overlaps this circle on the x axis (even if
// they were side by side), choose the y position at random. This
// makes the plot look a bit more natural and less like the Mandelbrot
// set!
item.y =
item.forbiddenY.head.next != item.forbiddenY.tail
? item.bestPosition
: this.radiusFun(item.datum) * (this.rng() - .5) * randomness2;
this._updateForbiddenYIntervals(item, all);
}
all.sort((a, b) => a.originalIndex - b.originalIndex);
return all.map(d => ({ datum: d.datum, x: d.x, y: d.y }));
}
// Random number generator (for reproducibility)
// https://stackoverflow.com/a/47593316
_sfc32(a, b, c, d) {
let rng = function() {
a >>>= 0;
b >>>= 0;
c >>>= 0;
d >>>= 0;
var t = (a + b) | 0;
a = b ^ (b >>> 9);
b = (c + (c << 3)) | 0;
c = (c << 21) | (c >>> 11);
d = (d + 1) | 0;
t = (t + d) | 0;
c = (c + t) | 0;
return (t >>> 0) / 4294967296;
};
for (let i = 0; i < 10; i++) {
rng();
}
return rng;
}
_updateForbiddenYIntervals(item, all) {
for (let step of [-1, 1]) {
let xDist;
let r = this.radiusFun(item.datum);
for (
let i = item.index + step;
i >= 0 &&
i < all.length &&
(xDist = Math.abs(item.x - all[i].x)) < r + this.maxR;
i += step
) {
let other = all[i];
if (other.placed) continue;
let sumOfRadii = r + this.radiusFun(other.datum);
if (xDist >= r + this.radiusFun(other.datum)) continue;
let yDist = Math.sqrt(sumOfRadii * sumOfRadii - xDist * xDist);
let forbiddenInterval = [item.y - yDist, item.y + yDist];
other.forbiddenY.addInterval(forbiddenInterval);
other.bestPosition = other.forbiddenY.bestAvailableValue();
}
}
}
}
class UnionOfIntervals {
constructor() {
this.head = { right: -Infinity };
this.tail = { left: Infinity };
this.head.next = this.tail;
this.tail.prev = this.head;
}
del(i) {
i.prev.next = i.next;
i.next.prev = i.prev;
}
insertBefore(interval, i) {
interval.prev = i.prev;
interval.next = i;
interval.prev.next = interval;
i.prev = interval;
}
addInterval(interval) {
let l = interval[0];
let r = interval[1];
interval = { left: l, right: r };
let nxt;
for (let i = this.head.next; i != this.tail; i = nxt) {
nxt = i.next;
// if `i` contains `interval`, do nothing
if (i.left <= l && i.right >= r) {
return;
}
// if `interval` contains `i`, delete `i`
if (l <= i.left && r >= i.right) {
this.del(i);
}
}
for (let i = this.head.next; i != this.tail; i = nxt) {
nxt = i.next;
if (r <= i.left) {
this.insertBefore(interval, i);
return;
}
if (r > i.left && r < i.right) {
i.left = l;
return;
}
if (l < i.right) {
if (r > i.next.left) {
// combine i and i.next
i.right = i.next.right;
this.del(i.next);
} else {
i.right = r;
}
return;
}
}
this.insertBefore(interval, this.tail);
}
bestAvailableValue() {
let best = Infinity;
let zeroAvailable = true;
for (let i = this.head.next; i != this.tail; i = i.next) {
if (i.left < 0 && i.right > 0) zeroAvailable = false;
if (Math.abs(i.left) < Math.abs(best)) best = i.left;
if (Math.abs(i.right) < Math.abs(best)) best = i.right;
}
return zeroAvailable ? 0 : best;
}
}
mobility_data = d3.csv("https://static01.nyt.com/newsgraphics/2020/03/27/coronavirus-behavior-int/d08526cd59a705b40c03e9950174af4a5bb72835/pct_change_04_01.csv")
One platform to build and deploy the best data apps
Experiment and prototype by building visualizations in live JavaScript notebooks. Collaborate with your team and decide which concepts to build out.
Use Observable Framework to build data apps locally. Use data loaders to build in any language or library, including Python, SQL, and R.
Seamlessly deploy to Observable. Test before you ship, use automatic deploy-on-commit, and ensure your projects are always up-to-date.