Found driving the Next UI against the sample app: the UI cannot send the
user's id, and core checks required fields and record locks before the
customizers run. An insert-only field behavior fills empty owner/user
fields with the session user (never on update, so an administrator's
partial edit cannot take ownership); naming another owner is still
refused.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Chi-square now pools the smallest category into Other one at a time until
every expected count is at least five (rows too, for independence), so a
crosstab with a single-record row still gets a valid test.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Extension 1.4.0: saved analyses of every kind re-run as the viewer
(fields they cannot analyze are dropped), owner-or-shared READ through
core record locks with owner-only writes (and a database CHECK so a
private row can never lose its read owner), personal stars, schedules
that run as the owner and mail only recipients who could run the analysis
themselves, CSV/XLSX export, the SQL Workbench handoff, an analysis
dashboard widget, and an Analyses home built from standard widgets.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Watches evaluate a metric per time bucket as their owner (SPIKE, DROP,
OUT_OF_RANGE, NEW_CATEGORY, NULL_RATE_CREEP with robust baselines) on a
hidden scheduled process, raise and resolve data.analysis.anomaly alerts
through QAlerts, and offer Evaluate now / Enable / Disable record actions
and a preview endpoint (extension 1.3.0). Pivot and crosstab labels are
human (Yes/No, Average Annual Salary). Contingent fields: a watch's saved
analysis depends on its table, table names use the core tables PVS, and
analyzable/excluded fields pick from the chosen table's fields.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Extension 1.2.0: compare two filters field by field with deltas and
significance (two-proportion z, Welch's t, chi-square with validity
rules and plain sentences), distribution shift over time with PSI, a
pushed-down Pearson correlation matrix with sampled scatter and trend,
and crosstabs with chi-square and Cramer's V. Pure-Java statistics
checked against scipy reference values; the privacy guard, field access
and drills apply as for profiles.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
POST /tables/{table}/pivot (extension 1.1.0): up to 3 row and 2 column
dimensions with date buckets or nice numeric bins, fields from to-one
exposed joins, count/sum/avg/min/max/distinct/percentile measures,
totals, Other-merging by top-N, the privacy guard and a drill per cell.
FieldAnalysisAccessResolver maps fields to FULL / EXACT_MATCH_ONLY /
NONE (core secured-data tiers when present, FieldSecurityLock otherwise);
only FULL fields are analyzable, drills never name others, and exact-run
cache keys include the user's field-access fingerprint.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every metric carries a plain-English label and a display-formatted value
(grouping, percent, the field's display format), histogram bins use nice
edges with formatted labels, date buckets are labelled by unit, warnings
appear only when true, and the Analyses home shows saved analyses,
watches and recent runs with honest empty states (and no mojibake).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
UI API extension data-analysis 1.0.0 (capabilities, column-profile,
distinct-values, table-profile, duplicates, runs) with typed
DataAnalysis*V1 schemas and the committed OpenAPI fragment. The analyze
permission is registered through core TableFeaturePermissionsMetaData.
CoreAnalysisBackend pushes percentiles, stddev, null/blank counts, numeric
and date buckets down through the RC.6 aggregate extensions; estimates
sample natively and scale counts, and the stand-in uses one primary-key
bound instead of an IN list. Builds against QQQ 4.1.0-RC.6.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
james.maesandClaude Opus 5.5