feat: mid-project lead scoring — generation framework

[shaypal5]

Summary

Adds the reusable generation and validation framework for the mid-project lead scoring dataset. No dataset files are committed here (leadforge is a public repo). The generated CSV and release docs live in leadforge-datasets-private.

Files added

File	Purpose
leadforge/pipelines/build_midproject.py	Pipeline module (seed=100, SUBSAMPLE_N=1200)
scripts/build_midproject_lead_scoring.py	Build CLI — python scripts/build_midproject_lead_scoring.py OUTPUT_DIR
scripts/validate_midproject_lead_scoring.py	Validation script with all spec checks
scripts/quick_baseline_eval_midproject.py	Multi-model baseline eval script

Dataset location

Dataset artifacts (CSV, docs, validation report) live in:
leadforge-datasets-private/lead_scoring_midproject/

Design

• Same schema and missingness patterns as lead_scoring_intro_v7.csv
• Different seed (100 vs v7's 42) → different rows, cannot copy class notebook outputs
• 1,200 rows (vs v7's 1,000), 30% conversion rate, snapshot day 20
• No __leakage__ columns
• Validated LR AUC=0.703, P@25=0.520 (Lift 1.73x), EV uplift +35% at K=25

Regenerate

python scripts/build_midproject_lead_scoring.py /path/to/output/
python scripts/validate_midproject_lead_scoring.py /path/to/output/lead_scoring_midproject.csv

Test plan

• No dataset files in branch history (force-pushed clean branch from main)
• Framework scripts are importable and lint-clean

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Adds a new “mid-project” variant of the lead-scoring teaching dataset (student-safe single CSV) plus the build/validate/eval tooling to reproduce and sanity-check it, aligned to the existing v7 schema/narrative.

Changes:

• Introduces lead_scoring_midproject.csv (1,200 rows) with accompanying BACKGROUND/RELEASE documentation and a saved validation report JSON.
• Adds a midproject pipeline module (leadforge/pipelines/build_midproject.py) and build/validate/baseline-eval scripts under scripts/.
• Updates .agent-plan.md with the completed work item entry for this dataset.

Reviewed changes

Copilot reviewed 8 out of 9 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
scripts/validate_midproject_lead_scoring.py	New validator for the midproject CSV (schema, missingness, baseline metrics, cohort/value-aware checks).
scripts/quick_baseline_eval_midproject.py	Quick baseline runner to reproduce headline metrics and feature importances.
scripts/build_midproject_lead_scoring.py	CLI to generate the midproject dataset CSV from the simulator/pipeline.
leadforge/pipelines/build_midproject.py	Midproject pipeline configuration (seed=100, subsample=1200) reusing common v6/v7 pipeline steps.
lead_scoring_midproject/validation_midproject_report.json	Persisted output from running the validator on the generated CSV.
lead_scoring_midproject/lead_scoring_midproject.csv	The generated student-safe dataset artifact.
lead_scoring_midproject/lead_scoring_midproject_RELEASE.md	Release documentation with metrics, teaching affordances, and provenance.
lead_scoring_midproject/lead_scoring_midproject_BACKGROUND.md	Student-facing narrative/background and feature descriptions.
.agent-plan.md	Tracks the completion of the midproject dataset work item.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

pr-agent-context report:

No unresolved review comments, failing checks, or actionable patch coverage gaps were found on PR #79 in repository https://github.com/leadforge-dev/leadforge. Treat this PR as all clear unless new signals appear.

Run metadata:

Tool ref: v4
Tool version: 4.0.21
Trigger: commit pushed
Workflow run: 25795848924 attempt 1
Comment timestamp: 2026-05-13T11:20:31.681850+00:00
PR head commit: cea57addb5cdf5127efd079e0f05aa79b0419453