Skip to content

feat: v4-M2 — snapshot day tuning + validation fix + release docs #24

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
shaypal5 merged 1 commit into from
Apr 29, 2026
Merged

feat: v4-M2 — snapshot day tuning + validation fix + release docs #24

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 11 additions & 9 deletions .agent-plan.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@

## Current System State

**v0.5.0 in progress — Milestones 7–11 complete, v4-M1 in PR.** Full simulation engine + render/bundle + exposure filtering + CLI commands + validation harness implemented. v4 engine changes + build pipeline ready. 609 tests passing.
**v0.5.0 in progress — Milestones 7–11 complete, v4 dataset shipped.** Full simulation engine + render/bundle + exposure filtering + CLI commands + validation harness implemented. v4 engine changes + build pipeline merged (PR #21). v4 dataset generated and validated. 609 tests passing.

---

Expand All @@ -24,7 +24,7 @@ See `docs/v4/design.md` for full details.
- [x] `scripts/spike_category_signal.py` — spike experiment validating category signal approach
- [x] Updated `CLAUDE.md`, `AGENTS.md`, `.agent-plan.md`

### v4-M1: Engine + build pipeline ✓ (in PR)
### v4-M1: Engine + build pipeline ✓ (PR #21, merged)

Engine changes:
- [x] Add `category_latent_correlations` to `difficulty_profiles.yaml` (intro profile, scale 1.8)
Expand All @@ -37,17 +37,19 @@ Engine changes:
- [x] 19 new tests (16 windowed snapshot + 3 category-latent correlations)

Build pipeline:
- [x] `scripts/build_v4_snapshot.py` — day-14 snapshot + leakage trap + structured missingness + subsampling
- [x] `scripts/build_v4_snapshot.py` — day-10 snapshot + leakage trap + structured missingness + subsampling
- [x] `scripts/validate_v4_dataset.py` — full validation per `docs/v4/validation_spec.md`
- [x] End-to-end: generate bundle → build CSV → validate → all 7 mandatory checks pass
- [x] LR AUC 0.659 (without trap); 0.03+ boost with trap
- [x] LR AUC 0.652 (without trap); 0.034 boost with trap

### v4-M2: Documentation + release ⬜
### v4-M2: Documentation + release ✓ (PR #22)

- [ ] Generate `lead_scoring_intro_v4.csv` (in datasets-private repo)
- [ ] Write `RELEASE_v4.md`
- [ ] Update dataset repo README
- [ ] Update `.agent-plan.md` to reflect completion
- [x] Generate `lead_scoring_intro_v4.csv` (in datasets-private repo)
- [x] Write `RELEASE_v4.md` with column dictionary, validation results, pedagogical notes
- [x] Update dataset repo README (v4 now recommended, v3 superseded)
- [x] Tune snapshot day (14 → 10) for leakage trap boost ≥ 0.03 after stacking fix
- [x] Add StandardScaler to validation script LR for convergence
- [x] Update `.agent-plan.md` to reflect completion

---

Expand Down
6 changes: 3 additions & 3 deletions scripts/build_v4_snapshot.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@
python scripts/build_v4_snapshot.py OUTPUT_CSV

Produces a 1000-row × 18-column CSV at ~30% conversion rate with:
- Day-14 windowed features
- Day-10 windowed features
- Structured missingness (MAR for web_sessions, seniority)
- Leakage trap (total_touches_all using full 90-day data)
- Stratified subsampling
Expand All @@ -27,7 +27,7 @@
# ---------------------------------------------------------------------------
SEED = 42
N_LEADS = 5000
SNAPSHOT_DAY = 14
SNAPSHOT_DAY = 10
SUBSAMPLE_N = 1000
TARGET_RATE = 0.30

Expand Down Expand Up @@ -72,7 +72,7 @@


def generate_bundle(seed: int = SEED, n_leads: int = N_LEADS) -> pd.DataFrame:
"""Generate a full bundle and return the day-14 snapshot."""
"""Generate a full bundle and return the day-10 snapshot."""
gen = Generator.from_recipe(
"b2b_saas_procurement_v1",
seed=seed,
Expand Down
9 changes: 6 additions & 3 deletions scripts/validate_v4_dataset.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,7 +16,7 @@
import pandas as pd
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import roc_auc_score
from sklearn.preprocessing import LabelEncoder
from sklearn.preprocessing import LabelEncoder, StandardScaler

# ---------------------------------------------------------------------------
# Constants
Expand Down Expand Up @@ -120,9 +120,12 @@ def _fit_lr(df: pd.DataFrame, exclude_cols: list[str] | None = None) -> float:
x_df = x_df.select_dtypes(include=[np.number])
x_df = x_df.fillna(x_df.median())

scaler = StandardScaler()
x_scaled = scaler.fit_transform(x_df)

lr = LogisticRegression(max_iter=2000, random_state=42)
lr.fit(x_df, y)
probs = lr.predict_proba(x_df)[:, 1]
lr.fit(x_scaled, y)
probs = lr.predict_proba(x_scaled)[:, 1]
return float(roc_auc_score(y, probs))


Expand Down
Loading