fix(notebook,#18144): aligner prose 0.007 -> 0.002 sur 2.11 LASSO conclusion c.27

[jsboige]

Grain: DEEP/notebook-python — lane myia-po-2026:CoursIA-2 — prev: DEEP/research-code #18152

Fix prose 0.007 -> 0.002 sur 2.11-Regularisation-Sparse-LASSO.ipynb (c.27)

Audit #18144 (campagne haiku lecture seule du 27/09, 2ᵉ passe) a releve une derive entre la prose de la cellule [27] (id c2d10co01) et la sortie de la cellule [15] (id c2d10col02) du carnet 2.11-Regularisation-Sparse-LASSO.ipynb :

• Prose : ElasticNet suit Lasso a +-0.007 pres
• Sortie : max abs = 0.002 (mesure sur 5 tirages, design c15)

La cellule [14] (id c2d10col01) du meme carnet dit d'ailleurs EN suit LASSO à ±0.002 près -- la cellule [27] contredit la cellule [14] qui presente pourtant le meme constat juste au-dessus.

Fix : remplacement de +-0.007 par +-0.002 dans la cellule [27]. Aucune cellule de code touchee, aucun re-execution due (C.2/C.3) : la sortie reste la meme, c'est la prose qui s'aligne sur la mesure.

Tranche single-notebook sur l'audit #18144 (42 constats sur 28 notebooks, relu par ai-01, claim par chemin pour cette lane). D'autres tranches sur d'autres carnets seront livrees par leur lane respective.

Verdict

CONFIRMED DRIFT -- le constat designe par #18144 sur ce carnet est reel, verifie cellule par cellule.

Invariant

• Aucune cellule de code modifiee (markdown uniquement)
• execution_count et outputs des cellules code inchanges (les seules sorties mesurees, [15], restent celles de main)
• Format JSON inchange (indent=1 + newline final preserve par aller-retour avant ecriture)

Gardes

Garde	Resultat
check_interp_positioning.py	findings total : 0
check_cell_source_parses.py (1 carnet)	findings: 0
Pre-commit (gitleaks, H.3 execution, #13326 compilabilite)	a passer au push

See #18144 -- la campagne reste ouverte, 41 constats non traites par cette lane.

🤖 Generated with Claude Code

[github-actions]

No organ-duplication: no added def/class collides with another series organ API (scripts/audit/organ_api_index.yaml).

Detector: python scripts/audit/detect_organ_duplication.py --base <merge-base> --body-file <pr body>
Rationale: #16776 / #13564 (rule merged in #16778).

[github-actions]

Golden-Set Execution (H.7 P3)

✅ 8/8 notebooks passed (certified reproducible)

Notebook	Status	Time
2.1-Workflow-ML.ipynb	✅ SUCCESS	3.8s
2.2-Descente-de-gradient.ipynb	✅ SUCCESS	9.0s
2.3-Regression-lineaire-logistique.ipynb	✅ SUCCESS	8.1s
2.4-Arbres-Forets-Ensembles.ipynb	✅ SUCCESS	3.7s
Search-01-StateSpace.ipynb	✅ SUCCESS	2.9s
SL-1-LogicalLearning.ipynb	✅ SUCCESS	1.9s
rl_4_multi_armed_bandits.ipynb	✅ SUCCESS	16.4s
GameTheory-04c-NashExistence-Python.ipynb	✅ SUCCESS	2.8s

Pinned lockfile: scripts/notebook_tools/golden_set.lock.txt (H.7 P3, axe A #4208)

[github-actions]

Notebook PR Validation: PASS

• Notebooks checked: 1
• Code cells validated: 10
• Result: All passed

Checks: H.1 (no errors), H.3 (execution_count), C.1 (no banned patterns)
Non-Python kernels (.NET/Lean): C.1 + errors only (execution_count advisory)
QuantConnect notebooks: C.1 + errors only (require QC Cloud for execution)

[github-actions]

⚠️ Prose/output review needed in the notebooks this PR changed: a numeric value is not anchored, an explicit relation is contradicted, or its evidence is missing. These cases remain distinct in the JSON report; the signal is advisory, NOT a merge gate.

Scope = notebooks CHANGED in this PR, not the whole corpus. Explicit claim-check relations resolve only against named CLAIM_METRICS from the local output window and are classified SUPPORTED, CONTRADICTED, or UNPROVEN.
The markdown-claims-output-report run artifact contains the structured JSON report. See python scripts/check_markdown_claims_output.py --help for re-running locally.
Detector rationale: c.290 / c.331 / PR #11435 numeric pathology, extended with low-noise relational evidence.

[github-actions]

Notebook outputs-required (H.4 schema): PASS (every code cell carries an outputs: list)

[clusterManager-Myia]

VERDICT: LGTM — alignement prose 0.007 → 0.002 dans la conclusion (c.27) de 2.11-Regularisation-Sparse-LASSO, au head 57569a85.

Vérifié (gate #17040 critère 2) : la valeur 0.002 est ancrée dans l'output committé de la cellule code 15 — ElasticNet suit LASSO a ±0.002 pres sur la feature 0 des paires — et la lecture markdown 16, placée immédiatement après ce code, la cite correctement ; la conclusion 27 la reprend en récapitulatif. Aucun 0.007 résiduel dans sources ni outputs du notebook entier. Le reste du diff est le reformatage de la cellule en ligne unique, sans changement de contenu.

[Hermes hermes-pr-review, cycle :00 28/09, host f6be46d1b7a3, sig=6a4f4281]

[jsboige]

[ADJOINT PREFLIGHT]
schema: 1
lane: myia-po-2026:CoursIA-3
pr: 18155
head: 57569a8
complete: true
body: read
comments-reviewed: 5
reviews-reviewed: 1
threads-reviewed: 0
threads-unresolved: 0
surfaces-sha256: b2602c412fc8d9c8829ca5abd03445ec4fad536b4bfe23573242db2ee07e614e
diff-files: 1
diff-additions: 1
diff-deletions: 19
checks: latest-wins-green
b0: clear
scope: pass
domain: pass
verdict: READY
[/ADJOINT PREFLIGHT]

Dossier READY sur PR #18155 (lane myia-po-2026:CoursIA-3, cycle c.250, 28/09 08:30Z).

Substance : fix(notebook,#18144): aligner prose 0.007 -> 0.002 sur 2.11 LASSO.

Substance : +1/-19 sur 1 fichier. Grain DEEP/notebook-python.

État : CLEAN+MERGEABLE+rev=APPROVED, B.0 rc=0, tous checks verts.

Grain: META/secretary -- lane myia-po-2026:CoursIA-3 -- prev: META/secretary c.249