Publication Readiness · IDA-PTW EEWS Review

TL;DR — Ringkas — The IDA-PTW framework has sound methodology and four independent experiments with reproducible, defensible results. However, the current manuscript draft is NOT yet publication-ready due to two critical gaps: (1) ~31 numeric claims across the Abstract, Tables 4, 7, 8, 11, 12, and 15 have no CSV artifact in the repository — they are quoted as scalars only; (2) the Stage 0 URPD, Stage 1 Intensity Gate, and Stage 1.5 Distance Regressor subsystems described in Sections IV.B–D are not implemented in any script found in the codebase. Framework IDA-PTW memiliki metodologi yang solid dan empat eksperimen independen dengan hasil yang dapat direproduksi dan dipertahankan. Namun, draft manuskrip saat ini BELUM siap publikasi karena dua gap kritis: (1) ~31 klaim numerik di Abstract, Tabel 4, 7, 8, 11, 12, dan 15 tidak memiliki artefak CSV di repositori — hanya dikutip sebagai skalar; (2) subsistem Stage 0 URPD, Stage 1 Intensity Gate, dan Stage 1.5 Distance Regressor yang dijelaskan di Section IV.B–D tidak diimplementasikan di skrip mana pun dalam codebase.

⬇ Download Full Report (.docx, 20 KB) ⬇ Unduh Laporan Lengkap (.docx, 20 KB) ⬇ Download Traceability Matrix (.csv) ⬇ Unduh Traceability Matrix (.csv)

Status Summary (83 manuscript claims)

Ringkasan Status (83 klaim manuskrip)

Status	Status	Count	Jumlah	Meaning	Arti
MATCH / FULL MATCH	MATCH / FULL MATCH	12	Exactly reproduced from authoritative CSV	Tereproduksi persis dari CSV otoritatif	OK
COMPUTE OK	3	Derived correctly from source CSV	Diturunkan dengan benar dari CSV sumber	OK
CLOSE	1	Within rounding tolerance; check acceptability	Dalam toleransi pembulatan; cek kelayakan	WARN
OFF / PARTIAL	OFF / PARTIAL	4	Value mismatches data; needs paper-side correction	Nilai tidak cocok; butuh koreksi di paper	WARN
NARR / UNCHECKED	NARR / UNCHECKED	3	Narrative claim; not tabulated/checked	Klaim narasi; belum tertabulasi/diperiksa	WARN
NO CSV	31	Claim has no reproducible CSV artifact in repo	Klaim tidak punya artefak CSV di repo	BLOCK
FABRICATED / WRONG / MISLABEL	FABRICATED / WRONG / MISLABEL	5	Number cannot be defended; must fix before submission	Angka tidak bisa dipertahankan; harus diperbaiki	BLOCK

Four-Phase Validation Results

Hasil Validasi Empat-Fase

1. INPUT — Dataset composition previously undisclosed

1. INPUT — Komposisi dataset belum didokumentasikan

The 25,058-trace dataset is a UNION of two sources: d1 (24,466 main Java-Sunda traces, PGA max ≈1.66 gal) and gb3 (592 injected traces from 3 Damaging events — Cianjur 2022-11-21, late-2023, 2024-04-27). This composition is not described in Section III.A and must be disclosed.

Dataset 25.058 trace adalah GABUNGAN dua sumber: d1 (24.466 trace utama Java-Sunda, PGA maks ≈1,66 gal) dan gb3 (592 trace injeksi dari 3 event Damaging — Cianjur 21-11-2022, late-2023, 27-04-2024). Komposisi ini belum dijelaskan di Section III.A dan harus didokumentasikan.

Parameter	Parameter	Manuscript	Manuskrip	Observed	Teramati	Action
Total traces	Total trace	25,058	25,058 ✓	Keep	Pertahankan
Distinct events	Event distinct	336	338	Correct to 338	Koreksi ke 338
IA-BMKG stations	Stasiun IA-BMKG	125	125 ✓	Keep	Pertahankan
Epicentral distance	Jarak epicentral	5–560 km	0.7–299.9 km (d1)	0,7–299,9 km (d1)	Correct both bounds	Koreksi kedua batas
Median distance	Jarak median	~109 km	121.9 km	121,9 km	Correct to ~122 km	Koreksi ke ~122 km
Weak class %	% kelas Weak	30,540 (94%)	24,466 (97.6%)	24.466 (97,6%)	Recompute all 3 bins	Hitung ulang 3 bin
Felt/Strong %	% Felt/Strong	1,931 (5.7%)	≈0 (d1)	Fabricated	Dibuat-buat
Damaging %	% Damaging	101 (0.3%)	≈592 gb3	Fabricated	Dibuat-buat

2. PROCESS — Three distinct runs mixed + missing subsystems

2. PROSES — Tiga run tercampur + subsistem hilang

PASS GroupKFold leakage check — 0 event overlap across 5 folds on N=2,747 stratified subset.
PASS Cek leakage GroupKFold — 0 overlap event di 5 fold pada subset stratified N=2.747.
FAIL Hyperparameter consistency — 4 different XGBoost configs across training scripts (RUN-A heavy 800/12, RUN-B fast 150/8, plus 2 intermediate). Source of 0.876 / 0.738 / 0.625 R² spread.
FAIL Konsistensi hiperparameter — 4 konfigurasi XGBoost berbeda di training script (RUN-A heavy 800/12, RUN-B fast 150/8, plus 2 konfig intermediate). Sumber spread R² 0,876 / 0,738 / 0,625.
FAIL Feature Dichotomy NOT implemented — keyword feature_weights appears 0 times in 6 training scripts. Manuscript Section IV.E claim is unsubstantiated.
FAIL Feature Dichotomy TIDAK diimplementasikan — keyword feature_weights muncul 0 kali di 6 skrip training. Klaim Section IV.E manuskrip tidak tersubstansiasi.
FAIL Stage 0 / 1 / 1.5 subsystems missing — No script for GradientBoostingClassifier, SMOTE, AUC computation, confusion matrix, or distance regressor. Tables 4, 7, 8 are unreproducible.
FAIL Subsistem Stage 0/1/1.5 tidak ada — Tidak ada skrip untuk GradientBoostingClassifier, SMOTE, komputasi AUC, matriks konfusi, atau regressor jarak. Tabel 4, 7, 8 tidak dapat direproduksi.
PASS Physics engine (scwfparam) validated — generate_golden_report.py uses log10 and sklearn r2_score correctly on 21,704-row golden subset.
PASS Engine fisis (scwfparam) tervalidasi — generate_golden_report.py menggunakan log10 dan sklearn r2_score dengan benar pada subset golden 21.704 baris.

3. OUTPUT — 46/46 numeric checks pass on authoritative CSVs

3. OUTPUT — 46/46 pemeriksaan angka lolos pada CSV otoritatif

All 14 CSV files in reports/analysis/ and reports/performance/ are machine-readable. A 46-point numeric re-verification passed 100%. These CSVs are publication-grade and correctly support Tables 13, 14, and the Fixed-PTW rows of Table 11.

Keempat belas file CSV di reports/analysis/ dan reports/performance/ semua readable. 46 pemeriksaan angka lolos 100%. CSV-CSV ini berkualitas publikasi dan mendukung Tabel 13, 14, dan baris Fixed-PTW Tabel 11 dengan benar.

4. TRACEABILITY — 31 NO-CSV claims + 5 defensible-defects

4. TRACEABILITY — 31 klaim NO-CSV + 5 defect fundamental

See the downloadable Traceability Matrix (.csv) for a claim-by-claim breakdown across 8 manuscript sections + 7 tables + abstract/conclusion. Each row maps a manuscript numeric claim to a CSV source and assigns a status.

Lihat Traceability Matrix (.csv) yang dapat diunduh untuk breakdown klaim-per-klaim di 8 section + 7 tabel + abstract/conclusion manuskrip. Setiap baris memetakan klaim numerik ke sumber CSV dan menetapkan status.

Priority Action Plan

Rencana Aksi Prioritas

🔴 BLOCKERS — must resolve before submission

🔴 BLOCKER — harus diselesaikan sebelum submit

Fix Table 11 labels: rename "Fixed 4s" → "Fixed 5s"; delete/regenerate "Fixed 6s" row. Effort: 5 minutes.
Perbaiki label Tabel 11: rename "Fixed 4s" → "Fixed 5s"; hapus/regenerasi baris "Fixed 6s". Waktu: 5 menit.
Correct Table 1 metadata (epicentral range, median distance, event count 336→338). Effort: 30 minutes.
Koreksi metadata Tabel 1 (range epicentral, jarak median, event count 336→338). Waktu: 30 menit.
Recompute Section III.C intensity distribution with documented d1/gb3 composition. Effort: 2 hours.
Hitung ulang distribusi intensitas Section III.C dengan dokumentasi komposisi d1/gb3. Waktu: 2 jam.
Produce Stage 0 URPD + Stage 1 + Stage 1.5 metrics CSVs (or trim claims). Effort: 1.5–6 days total.
Produksi CSV metrik Stage 0 URPD + Stage 1 + Stage 1.5 (atau trim klaim). Waktu: 1,5–6 hari total.
Produce sigma decomposition CSV (Table 15 τ/φ/σ per period). Effort: 1 day.
Produksi CSV dekomposisi sigma (Tabel 15 τ/φ/σ per-periode). Waktu: 1 hari.
Re-run Full-Wave (~341 s) + Total MiniSEED (~430 s) with retained CSVs, or drop Evidence B. Effort: 1–2 days.
Re-run Full-Wave (~341 s) + Total MiniSEED (~430 s) dengan CSV tersimpan, atau drop Evidence B. Waktu: 1–2 hari.
Produce end-to-end IDA-PTW operational CSV (R² = 0.7309 currently has no source). Effort: 1 day.
Produksi CSV end-to-end IDA-PTW operational (R² = 0,7309 saat ini tidak punya sumber). Waktu: 1 hari.

Two Paths Forward

Dua Jalur ke Depan

Full scope (preserves manuscript ambition) — 2–4 weeks Scope penuh (pertahankan ambisi manuskrip) — 2–4 minggu

Retrieve/re-run legacy Full-Wave c7a50193 and Total MiniSEED c3399cac experiments; implement Stage 0/1/1.5 subsystems with CSV outputs; produce sigma decomposition CSV; fix Table 1 metadata and Table 11 labels; resubmit.

Retrieve/re-run eksperimen legacy Full-Wave c7a50193 dan Total MiniSEED c3399cac; implementasi subsistem Stage 0/1/1.5 dengan output CSV; produksi CSV dekomposisi sigma; perbaiki metadata Tabel 1 dan label Tabel 11; submit ulang.

Reduced scope (fastest path to publication) — 3–5 days Scope reduksi (jalur tercepat ke publikasi) — 3–5 hari

Restructure the paper around what is empirically verified: Evidence A (Fixed PTW benchmark), C (saturation), D (P-arrival robustness), plus stratified IDA-PTW 103-period result. Defer Stage 0/1/1.5 + sigma decomposition + Full-Wave ceiling to a follow-up paper.

Restructure paper di sekitar yang empirically verified: Evidence A (benchmark Fixed PTW), C (saturation), D (robustness P-arrival), plus hasil IDA-PTW 103-periode stratified. Defer Stage 0/1/1.5 + dekomposisi sigma + Full-Wave ceiling ke paper lanjutan.

Sections Ready to Submit AS-IS

Section yang Siap Submit Apa Adanya

Section V.C / Table 13 — Magnitude saturation test (6 values × N=1,204 match CSV exactly).
Section V.C / Tabel 13 — Uji magnitude saturation (6 nilai × N=1.204 match CSV persis).
Section V.D / Table 14 — P-arrival sensitivity (21 values × 7 shifts × 3 periods — all match CSV exactly).
Section V.D / Tabel 14 — Sensitivitas P-arrival (21 nilai × 7 shift × 3 periode — semua match CSV persis).
Section V.A Table 11 rows for Fixed 2/3/8 s — match benchmark_results_fixed.csv exactly.
Section V.A Tabel 11 baris Fixed 2/3/8 s — match benchmark_results_fixed.csv persis.
Section III.B QC filter counts — no numeric issues.
Section III.B QC filter counts — tidak ada masalah numerik.
Reference list (89 citations) — complete and internally consistent.
Reference list (89 sitasi) — lengkap dan konsisten internal.

Report Preview

Pratinjau Laporan

Page 1 — Executive Verdict + Status Summary

Halaman 1 — Executive Verdict + Status Summary

Page 4 — Traceability matrix (excerpt) + Priority Action Plan

Halaman 4 — Traceability matrix (cuplikan) + Rencana Aksi Prioritas

Report generated 2026-04-22 · 83 manuscript claims cross-verified · 46 CSV numeric checks (100% pass)Laporan dihasilkan 22-04-2026 · 83 klaim manuskrip di-cross-verify · 46 pemeriksaan angka CSV (100% lolos)

Publication Readiness Report

Laporan Kesiapan Publikasi