Skip to content

Board updates (rolling) — merge whenever - #14

Closed
ahrzb wants to merge 1 commit into
masterfrom
pm-board
Closed

Board updates (rolling) — merge whenever#14
ahrzb wants to merge 1 commit into
masterfrom
pm-board

Conversation

@ahrzb

@ahrzb ahrzb commented Jul 23, 2026

Copy link
Copy Markdown
Owner

Standing PR for board changes. Board-only (backlog/**) — never code.

I push planning work here as it happens instead of opening a PR per change. Merge it whenever suits you; there's no urgency on any single commit and nothing here blocks a dev. After you merge, I'll open a fresh one from the same branch for the next batch.

What lands here: new tickets and drafts, status/assignee changes, priority and dependency edits, ticket descriptions, verification notes when I close something.

If anything in here is contentious it'll say so in the ticket body rather than sitting in the PR description, so the reasoning stays with the ticket after merge.


Currently included

DRAFT-14 — per-group transformer fitting, transformer(x) OVER (PARTITION BY g)

AmirHossein's proposal, routed via Wren from TASK-3 brainstorming. Wren's measurements are recorded verbatim in the ticket since they're what make it scopeable.

The reframe that matters: a transformer ref is fitted once globally today — measured, country a={10,20}, b={30,50} gives avg_age=[27.5], not per-country 15/40. That's the correct semantic (matches sklearn, where a Pipeline step fits once on all training data), so this is a new feature, not a bug.

The split is the spine of the ticket:

  • (a) SQL-expressible (StandardScaler, MinMax, MaxAbs, mean-SimpleImputer) — already works today. Pure desugar, zero engine work. Measured: PARTITION BY yields one state row per group, batch == infer, unseen groups return NULL for free.
  • (b) Opaque sklearn objects — genuinely new. N fitted clones keyed by group, per-group state through the artifact and lookup path.

Four questions need your ruling before it becomes a task — unseen-group policy for (b), shorthand vs standard PARTITION BY syntax, whether (a) ships alone first, and artifact size at high group cardinality.

Flagged as distinct from DRAFT-11: that one is argument binding, this is what data a transformer is fitted on.

🤖 Generated with Claude Code

@claude-agent-ahrzb

Copy link
Copy Markdown

Recreating this fresh at AmirHossein's request — same pm-board branch and same contents, new PR.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant