The datasets

    Terms used on this page

    Calibration fit

    Every level of this build's target surface

    Within 10 % and within 25 % by family, each level separately

    National rows (country level)

    Region rows

    Constituency rows

    Local authority rows

    The worst-fitting national rows

    The worst-fitting local cells

    What the sparse refit lost against the dense solve

    Area support and weights

    Effective sample size by constituency

    Effective sample size by local authority

    The thinnest areas

    Weights

    Size selection

    Budget search

    Sparse refit loss trajectory

    Gates

    Downstream checks

    What an area-level analysis would rest on

    Out of sample: the ONS small-area income estimates by local authority

    National scorecard, one engine

    The poverty pair at HBAI's year

    Admin anchors (T4)

    Reform replays (T5)

    Variable distributions (T3)

    Dashboard replays (T6)

    Footprint

    Compared with the enhanced FRS

    National rows, by source family

    Constituency rows, by target family

    Local authority rows, by target family

    The same comparison as a table

    Each dataset scored on each surface

    Local families: where each side's targets come from

    Local families: how far apart the two target values are

    National rows on both surfaces, by family

    Rows on this build's surface with no incumbent counterpart

    Incumbent rows this build does not bind, and why

    Not measured

    What this page cannot say

      Sources and files

      Where the targets, floors and numbers come from

      Where the numbers live