How it works › Explain the data

Zach, on the actuator-models views: "with data-ui as the base it's really about making these dashboards that really explain the data." A view can open with its answer and make its numbers readable, without a second page.

- type: pivot
  name: Benches
  intro:
    text: |
      **How does every model do on every bench?** A row is a model, rolled up across the actuators it ran on; a
      column is one bench test; a cell is the median error of every run behind it, in multiples of that actuator's
      measurement floor (1 × = as close as the rack repeats itself; twin-grade is ≤ 1.5 ×). In the strip, each dot
      is a run and the tall tick their median, on one scale per column (its ends in the header). **Roll up** in the
      toolbar brings the actuators back; click a cell for its runs, a model's name for its page.
    kpi:
      filters:
        and:
          - 'kind == "run"'
          - '!x_floor.isEmpty()'
      groupBy: { property: bench }
      kpi:
        value: { property: x_floor, summary: Median }
        gate: { max: 1.5 }
  filters:
    and:
      - 'kind == "run"'
      - '!x_floor.isEmpty()'
  order: [variant, actuator]
  groupBy: { property: rung, direction: ASC }
  splitBy: [bench, test]
  value: { property: x_floor, summary: Median, display: dist }
  rollup: { keep: [variant], dims: [actuator, variant, model, rung, family] }
  columnSize: { variant: 200 }

The keys

• intro: {text, kpi} on any view: text is markdown (headings, lists, code, tables, links); kpi is an inline type: kpi entry over the same data and rail filter. Drawn above the status line.
• rollup: {keep, agg, dims} on a table or pivot: one row per combination of the kept dimensions, the rest aggregated. A number becomes a cell with its median (or agg: mean, min, max, count, spread, sum), its range and its count; text that disagrees reads "N values", never a made-up value. Click a rolled row for the rows behind it.
• columnGroups: {label: [properties]} on a table: a header over its columns. A pivot's column groups draw the same way.
• Displays (display: {p: …}): stats splits a property into sub-columns (median, min, max, n, runs; or value, low, high, basis); range puts low–high beside the value; value shows the basis mark with the range on hover; dist draws every value on one scale down the column, ends in the header; markdown renders text as a document in the peek.
• Fixed slots. A number cell is [number][range][mark][count] with the column's slot widths, so digits and marks line up down a column whatever the value.
• Spread (max − min) joins Bases' summaries, and lowercase names (median, mean, …) work wherever a summary does.

- type: pivot
  name: Benches
  intro:
    text: |
      **How does every model do on every bench?** A row is a model, rolled up across the actuators it ran on; a
      column is one bench test; a cell is the median error of every run behind it, in multiples of that actuator's
      measurement floor (1 × = as close as the rack repeats itself; twin-grade is ≤ 1.5 ×). In the strip, each dot
      is a run and the tall tick their median, on one scale per column (its ends in the header). **Roll up** in the
      toolbar brings the actuators back; click a cell for its runs, a model's name for its page.
    kpi:
      filters:
        and:
          - 'kind == "run"'
          - '!x_floor.isEmpty()'
      groupBy: { property: bench }
      kpi:
        value: { property: x_floor, summary: Median }
        gate: { max: 1.5 }
  filters:
    and:
      - 'kind == "run"'
      - '!x_floor.isEmpty()'
  order: [variant, actuator]
  groupBy: { property: rung, direction: ASC }
  splitBy: [bench, test]
  value: { property: x_floor, summary: Median, display: dist }
  rollup: { keep: [variant], dims: [actuator, variant, model, rung, family] }
  columnSize: { variant: 200 }

Comes from docs/CONTRACT.md §16 (explain the data) in docs/.