Central Intake Data Validation

BonQuery computed values vs. City of Toronto published figures

Validation of BonQuery’s Central Intake call statistics against the City of Toronto’s published Shelter System Requests for Referrals table, plus a monthly estimate of the Bridging and Triage population absent from the CKAN export.

Author

Miriam Marling

This page validates BonQuery’s computed monthly averages for the three Central Intake metrics against the City of Toronto’s published Shelter System Requests for Referrals table (reference snapshot: 2026-05-25, covering April 2024 – April 2026). It also quantifies the monthly gap between BonQuery’s CKAN-based nightly occupancy and the City’s published figure — the gap represents the Bridging and Triage programs that are tracked in the City’s operational system but not published to open data.

Validation: Calls referred, Unmatched callers, Calls handled

Tolerance applied: ±0.1 for Referred and Unmatched (City publishes one decimal place); ±1 for Handled (City publishes integers).

ci = FileAttachment("../data/central_intake.json").json()

cityRef = [
  {year:2024, month:4,  referred:5.7,  unmatched:224.5, handled:991,  occupancy:9734},
  {year:2024, month:5,  referred:7.3,  unmatched:220.2, handled:993,  occupancy:9936},
  {year:2024, month:6,  referred:5.2,  unmatched:231.1, handled:955,  occupancy:9850},
  {year:2024, month:7,  referred:6.2,  unmatched:235.7, handled:978,  occupancy:9815},
  {year:2024, month:8,  referred:7.3,  unmatched:233.0, handled:987,  occupancy:9713},
  {year:2024, month:9,  referred:7.1,  unmatched:223.0, handled:936,  occupancy:9615},
  {year:2024, month:10, referred:5.2,  unmatched:225.1, handled:929,  occupancy:9603},
  {year:2024, month:11, referred:16.0, unmatched:191.6, handled:799,  occupancy:9623},
  {year:2024, month:12, referred:25.9, unmatched:114.1, handled:604,  occupancy:9911},
  {year:2025, month:1,  referred:38.3, unmatched:89.0,  handled:577,  occupancy:10228},
  {year:2025, month:2,  referred:32.4, unmatched:101.0, handled:553,  occupancy:10150},
  {year:2025, month:3,  referred:22.6, unmatched:135.9, handled:628,  occupancy:9778},
  {year:2025, month:4,  referred:9.9,  unmatched:156.6, handled:708,  occupancy:9447},
  {year:2025, month:5,  referred:12.7, unmatched:148.7, handled:614,  occupancy:9181},
  {year:2025, month:6,  referred:15.8, unmatched:117.3, handled:509,  occupancy:9065},
  {year:2025, month:7,  referred:22.5, unmatched:95.7,  handled:500,  occupancy:8900},
  {year:2025, month:8,  referred:16.1, unmatched:119.8, handled:561,  occupancy:8765},
  {year:2025, month:9,  referred:13.9, unmatched:148.5, handled:627,  occupancy:8649},
  {year:2025, month:10, referred:12.9, unmatched:154.8, handled:653,  occupancy:8697},
  {year:2025, month:11, referred:21.7, unmatched:122.2, handled:596,  occupancy:8742},
  {year:2025, month:12, referred:42.4, unmatched:39.4,  handled:414,  occupancy:9035},
  {year:2026, month:1,  referred:55.1, unmatched:30.4,  handled:440,  occupancy:9217},
  {year:2026, month:2,  referred:39.3, unmatched:32.8,  handled:421,  occupancy:9206},
  {year:2026, month:3,  referred:35.4, unmatched:50.5,  handled:446,  occupancy:8742},
  {year:2026, month:4,  referred:20.0, unmatched:85.6,  handled:496,  occupancy:8391}
]

MONTH_ABB = ["","Jan","Feb","Mar","Apr","May","Jun",
             "Jul","Aug","Sep","Oct","Nov","Dec"]

// Index BonQuery monthly data by year+month for O(1) lookup
bqIndex = {
  const idx = new Map()
  ci.monthly.forEach(r => idx.set(r.year * 100 + r.month, r))
  return idx
}

// Pass/fail helpers
passRef  = (bq, city) => bq != null && Math.abs(bq - city) <= 0.10
passUnm  = (bq, city) => bq != null && Math.abs(bq - city) <= 0.10
passHdl  = (bq, city) => bq != null && Math.abs(bq - city) <= 1.0

// Rendered pass/fail mark
mark = (pass, bqVal) =>
  bqVal == null
    ? `<span style="color:var(--bq-fg-disabled)">—</span>`
    : pass
      ? `<span style="color:var(--bq-success);font-weight:700">✓</span>`
      : `<span style="color:var(--bq-danger);font-weight:700">✗</span>`

// Assemble one row of the validation table
function valRow(c, i) {
  const bq  = bqIndex.get(c.year * 100 + c.month)
  const br  = bq ? bq.referred_mean  : null
  const bu  = bq ? bq.unmatched_mean : null
  const bh  = bq ? bq.handled_mean   : null
  const pr  = passRef(br, c.referred)
  const pu  = passUnm(bu, c.unmatched)
  const ph  = passHdl(bh, c.handled)

  const bg = i % 2 === 0 ? "var(--bq-bg)" : "var(--bq-bg-alt)"
  const td = `padding:5px 9px;text-align:center;border-bottom:1px solid var(--bq-border)`
  const tdb = td + ";border-left:2px solid var(--bq-chart-blue)"

  return html`<tr style="background:${bg}">
    <td style="${td}">${c.year}</td>
    <td style="${td}">${MONTH_ABB[c.month]}</td>
    <td style="${tdb}">${br != null ? br.toFixed(1) : "—"}</td>
    <td style="${td}">${c.referred.toFixed(1)}</td>
    <td style="${td}">${mark(pr, br)}</td>
    <td style="${tdb}">${bu != null ? bu.toFixed(1) : "—"}</td>
    <td style="${td}">${c.unmatched.toFixed(1)}</td>
    <td style="${td}">${mark(pu, bu)}</td>
    <td style="${tdb}">${bh != null ? Math.round(bh).toLocaleString() : "—"}</td>
    <td style="${td}">${c.handled.toLocaleString()}</td>
    <td style="${td}">${mark(ph, bh)}</td>
  </tr>`
}

// Count pass/fails across all 25 months
valSummary = {
  let ref_p=0, unm_p=0, hdl_p=0, ref_f=0, unm_f=0, hdl_f=0
  cityRef.forEach(c => {
    const bq = bqIndex.get(c.year * 100 + c.month)
    const br = bq ? bq.referred_mean  : null
    const bu = bq ? bq.unmatched_mean : null
    const bh = bq ? bq.handled_mean   : null
    if (br != null) { passRef(br,c.referred) ? ref_p++ : ref_f++ }
    if (bu != null) { passUnm(bu,c.unmatched) ? unm_p++ : unm_f++ }
    if (bh != null) { passHdl(bh,c.handled)  ? hdl_p++ : hdl_f++ }
  })
  return {ref_p, ref_f, unm_p, unm_f, hdl_p, hdl_f}
}

// Summary counts banner — rendered before the table
html`<p style="font-size:0.9em;margin:0 0 10px">
  <strong>Pass / fail (25 months):</strong>
  Referred ${valSummary.ref_p}/25 ✓
  &nbsp;·&nbsp;
  Unmatched ${valSummary.unm_p}/25 ✓
  &nbsp;·&nbsp;
  Handled ${valSummary.hdl_p}/25 ✓
</p>`

// Validation table — one row per month, three metric triplets (BQ | City | ✓✗).
// Background/color applied per-<th> to survive Quarto Bootstrap's <thead> overrides.
{
  const H = `background:var(--bq-th-bg);color:var(--bq-th-fg);font-weight:600;font-size:0.82em;padding:6px 9px;text-align:center`
  const HB = H + ";border-left:2px solid var(--bq-chart-blue)"

  return html`<div style="overflow-x:auto">
<table style="width:100%;border-collapse:collapse;font-size:0.88em">
  <thead>
    <tr>
      <th style="${H}">Year</th>
      <th style="${H}">Month</th>
      <th style="${HB}">Referred BQ</th>
      <th style="${H}">Referred City</th>
      <th style="${H}"> </th>
      <th style="${HB}">Unmatched BQ</th>
      <th style="${H}">Unmatched City</th>
      <th style="${H}"> </th>
      <th style="${HB}">Handled BQ</th>
      <th style="${H}">Handled City</th>
      <th style="${H}"> </th>
    </tr>
  </thead>
  <tbody>
    ${cityRef.map((c, i) => valRow(c, i))}
  </tbody>
</table>
</div>`
}

Failures

// List only the rows that failed at least one metric
{
  const failures = cityRef.flatMap(c => {
    const bq = bqIndex.get(c.year * 100 + c.month)
    const br = bq ? bq.referred_mean  : null
    const bu = bq ? bq.unmatched_mean : null
    const bh = bq ? bq.handled_mean   : null
    const out = []
    if (br != null && !passRef(br, c.referred))
      out.push(`${MONTH_ABB[c.month]} ${c.year} — Referred: BonQuery ${br.toFixed(1)} vs City ${c.referred.toFixed(1)} (diff ${(br - c.referred).toFixed(2)})`)
    if (bu != null && !passUnm(bu, c.unmatched))
      out.push(`${MONTH_ABB[c.month]} ${c.year} — Unmatched: BonQuery ${bu.toFixed(1)} vs City ${c.unmatched.toFixed(1)} (diff ${(bu - c.unmatched).toFixed(2)})`)
    if (bh != null && !passHdl(bh, c.handled))
      out.push(`${MONTH_ABB[c.month]} ${c.year} — Handled: BonQuery ${Math.round(bh).toLocaleString()} vs City ${c.handled.toLocaleString()} (diff ${(bh - c.handled).toFixed(1)})`)
    return out
  })

  if (failures.length === 0)
    return html`<p style="color:var(--bq-success)">All cells pass.</p>`

  return html`<ul style="font-size:0.9em">
    ${failures.map(f => html`<li>${f}</li>`)}
  </ul>`
}

May 2024: Total calls handled. BonQuery’s computed monthly average (976.6 calls/day) falls 16.4 calls/day below the City’s published value (993). BonQuery’s extract covers all 31 days with no zeros or nulls, ruling out missing data as the cause. This points to a scope difference in the CKAN extract for that month — possibly a program or call type the City includes in its published figure but which is absent from the open data resource. All other months match within ±0.5, consistent with the City rounding to integers.

Nov 2024: Unmatched callers. BonQuery 191.4 vs City 191.6 (diff −0.2). Just outside the ±0.1 tolerance; likely a rounding difference in how the City computes its published monthly average.

Estimated Bridging & Triage population (not in CKAN export)

The City’s published nightly occupancy figure includes programs tracked in its operational system that are not published to open data (Bridging and Triage programs). BonQuery’s occupancy is computed from the CKAN export only. The gap below — City minus BonQuery — is an estimate of the Bridging and Triage population for each month. It is not a validation failure; it is an expected structural difference between the two data sources.

// Occupancy gap table — City published vs BonQuery CKAN, with the gap highlighted
{
  const H  = `background:var(--bq-th-bg);color:var(--bq-th-fg);font-weight:600;font-size:0.82em;padding:6px 9px;text-align:center`

  return html`<div style="overflow-x:auto">
<table style="width:100%;border-collapse:collapse;font-size:0.88em">
  <thead>
    <tr>
      <th style="${H}">Year</th>
      <th style="${H}">Month</th>
      <th style="${H}">BonQuery (CKAN)</th>
      <th style="${H}">City published</th>
      <th style="${H}">Gap (City − BonQuery)</th>
    </tr>
  </thead>
  <tbody>
    ${cityRef.map((c, i) => {
      const bq  = bqIndex.get(c.year * 100 + c.month)
      const bo  = bq ? bq.occupancy_mean : null
      const gap = bo != null ? c.occupancy - bo : null

      const bg = i % 2 === 0 ? "var(--bq-bg)" : "var(--bq-bg-alt)"
      const td = `padding:5px 9px;text-align:center;border-bottom:1px solid var(--bq-border)`

      const isDark = document.body.classList.contains("quarto-dark")
      const gapStyle = gap == null ? td
        : gap > 200 ? td + `;background:${isDark ? "rgba(217,119,6,0.25)" : "#ffe4cc"};font-weight:700`
        : gap > 80  ? td + `;background:${isDark ? "rgba(217,119,6,0.12)" : "#fff3e6"}`
        : td

      return html`<tr style="background:${bg}">
        <td style="${td}">${c.year}</td>
        <td style="${td}">${MONTH_ABB[c.month]}</td>
        <td style="${td}">${bo != null ? bo.toFixed(1) : "—"}</td>
        <td style="${td}">${c.occupancy.toLocaleString()}</td>
        <td style="${gapStyle}">${gap != null ? gap.toFixed(1) : "—"}</td>
      </tr>`
    })}
  </tbody>
</table>
</div>`
}

Larger gaps in late 2024 through early 2025 coincide with the peak of the winter shelter emergency response, when Bridging and Triage programs are likely operating at higher capacity. Smaller gaps in summer months are consistent with reduced emergency-response activity.

Data source

All data comes from the City of Toronto’s Central Intake Calls dataset and the Daily Shelter & Overnight Service Occupancy & Capacity dataset, published on the City’s open data portal. City reference values are from the Shelter System Requests for Referrals page, snapshot 2026-05-25.

Contains information licensed under the Open Government Licence – Toronto.