Twenty years of job ads, from the official count

requireLocal = require.alias({
  "d3": "/libs/d3-7.8.5.min.js",
  "d3@7.8.5/dist/d3.min.js": "/libs/d3-7.8.5.min.js",
  "htl": "/libs/htl-0.3.1.min.js"
})
d3 = requireLocal("d3")
htl = requireLocal("htl")
Plot = requireLocal("/libs/plot-0.6.11.umd.min.js")

ivi = FileAttachment("data/ivi_aust.csv").csv({typed: true})
cambios = FileAttachment("data/ivi_cambios.csv").csv({typed: true})
fmt = d3.format(",")
mes = d => d.toLocaleDateString("en-AU", {month: "long", year: "numeric", timeZone: "UTC"})
base2019 = d3.rollup(ivi.filter(d => d.mes.getUTCFullYear() === 2019), v => d3.mean(v, d => d.avisos), d => d.codigo)
grupos = ivi.filter(d => d.codigo !== 0).map(d => ({...d, indice: 100 * d.avisos / base2019.get(d.codigo)}))
total = ivi.filter(d => d.codigo === 0).map(d => ({...d, indice: 100 * d.avisos / base2019.get(0)}))
ultimoMes = d3.max(ivi, d => d.mes)
ba = cambios.find(d => d.codigo === 2611)
tot = cambios.find(d => d.codigo === 0)
// Sin signo menos: la web no lleva guiones (Jona, 16/10/2026)
flecha = v => v < 0 ? `▼ ${Math.abs(v)}%` : v > 0 ? `▲ ${v}%` : "0%"
sentido = v => v < 0 ? "baja" : v > 0 ? "sube" : ""
// El nucleo de la pagina en rosa, el total en negro, los demas en gris
ROSA = "#DB2777"
TINTA = "#0F172A"

html`<p>In ${mes(ultimoMes)}, ads for <strong>ICT Business and Systems Analysts</strong> (the official
group closest to data and business analysts) were <strong>${Math.abs(ba.pct_vs_2022)}% ${ba.pct_vs_2022 < 0 ? "below" : "above"}</strong>
their 2022 average (${fmt(ba.ultimo)} a month against ${fmt(ba.prom_2022)}). All job ads in
Australia were ${Math.abs(tot.pct_vs_2022)}% ${tot.pct_vs_2022 < 0 ? "below" : "above"} 2022 over the same time:
the data and tech groups fell much harder than the market as a whole.</p>`

Analysts against the whole market

Each line is rescaled so that its 2019 average is 100. Pink: ICT Business and Systems Analysts. Black: every job ad in Australia. Grey: the other data and tech groups in the table below (hover a line to see which).

destacado = grupos.filter(d => d.codigo === 2611)
otros = grupos.filter(d => d.codigo !== 2611)
fin = arr => arr.filter(d => +d.mes === +ultimoMes)
Plot.plot({
  height: 440,
  marginLeft: 50,
  marginRight: 185,
  x: {label: null},
  y: {label: "Job ads per month (2019 average = 100)", grid: true},
  marks: [
    Plot.ruleY([100], {stroke: "#94a3b8"}),
    Plot.lineY(otros, {x: "mes", y: "indice", z: "titulo", stroke: "#cbd5e1", strokeWidth: 1, tip: true,
      title: d => `${d.titulo}\n${mes(d.mes)}: ${fmt(Math.round(d.avisos))} ads a month (${Math.round(d.indice)} vs 2019)`}),
    Plot.lineY(total, {x: "mes", y: "indice", stroke: TINTA, strokeWidth: 2, tip: true,
      title: d => `All job ads in Australia\n${mes(d.mes)}: ${fmt(Math.round(d.avisos))} ads a month (${Math.round(d.indice)} vs 2019)`}),
    Plot.lineY(destacado, {x: "mes", y: "indice", stroke: ROSA, strokeWidth: 2.5, tip: true,
      title: d => `ICT Business and Systems Analysts\n${mes(d.mes)}: ${fmt(Math.round(d.avisos))} ads a month (${Math.round(d.indice)} vs 2019)`}),
    Plot.text(fin(destacado), {x: "mes", y: "indice", text: () => "Business & systems analysts", dx: 6, textAnchor: "start", fill: ROSA, fontWeight: 600}),
    Plot.text(fin(total), {x: "mes", y: "indice", text: () => "All job ads", dx: 6, textAnchor: "start", fill: TINTA})
  ]
})

The numbers

html`<table class="tabla-simple">
  <thead><tr><th>Official group</th><th>2019 avg</th><th>2022 avg</th><th>${mes(ultimoMes)}</th><th>vs 2022</th><th>vs a year before</th></tr></thead>
  <tbody>${cambios.slice().sort((a, b) => a.codigo - b.codigo).map(d => html`<tr>
    <td>${d.codigo === 0 ? "All job ads in Australia" : d.titulo}</td>
    <td class="n">${fmt(d.prom_2019)}</td><td class="n">${fmt(d.prom_2022)}</td><td class="n">${fmt(d.ultimo)}</td>
    <td class="n ${sentido(d.pct_vs_2022)}">${flecha(d.pct_vs_2022)}</td>
    <td class="n ${sentido(d.pct_vs_hace_un_ano)}">${flecha(d.pct_vs_hace_un_ano)}</td>
  </tr>`)}</tbody>
</table>`

There is no “data analyst” line in the official classification (ANZSCO): data analysts are spread across these groups. At this level of detail each value is an average of three months. Source: Internet Vacancy Index, Jobs and Skills Australia.