Improving Tsunami Prediction Accuracy Through Stacking Ensemble and Machine Learning Data Optimization

Hendi Firmansyah, Hendi Firmansyah, Ingrid Nurtanio, Intan Sari Areni, Joshua Purba

2025

Abstract

Accurate tsunami prediction is critical for disaster risk reduction. We propose a stacking ensemble that combines Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM), K‑Nearest Neighbors (KNN), Artificial Neural Network (ANN), Naïve Bayes (NB), and Gradient Boosting (GB) with a Logistic Regression meta‑learner. Global tectonic and volcanic datasets from BMKG, NOAA, and USGS are unified under a domain‑aware pipeline (imputation, normalization, outlier mitigation, and imbalance handling). The model outperforms single classifiers, achieving ROC‑AUC 90.1% (tectonic) with accuracy 84.2%, and ROC‑AUC 87.6% (volcanic) with accuracy 85.8%. Feature‑level signals align with geophysical intuition: magnitude, depth, and subduction status dominate tectonic cases, while VEI and volcano elevation are most informative for volcanic events. The limitation is a modest volcanic recall of 59.3%. Ablation indicates SMOTE improves volcanic recall but not tectonic, highlighting domain‑specific effects of oversampling. The contribution lies in a unified, deployable pipeline spanning tectonic–volcanic sources that generalizes across heterogeneous data and supports near‑real‑time operation on commodity hardware. Future work includes external validation, threshold calibration, and explainability integration.

Paper Citation

in Harvard Style

Firmansyah H., Nurtanio I., Areni I. and Purba J. (2025). Improving Tsunami Prediction Accuracy Through Stacking Ensemble and Machine Learning Data Optimization. In Proceedings of the 1st International Conference on Research and Innovations in Information and Engineering Technology - Volume 1: RITECH; ISBN 978-989-758-784-9, SciTePress, pages 224-230. DOI: 10.5220/0014276300004928

in Bibtex Style

@conference{ritech25,
author={Hendi Firmansyah and Ingrid Nurtanio and Intan Sari Areni and Joshua Purba},
title={Improving Tsunami Prediction Accuracy Through Stacking Ensemble and Machine Learning Data Optimization},
booktitle={Proceedings of the 1st International Conference on Research and Innovations in Information and Engineering Technology - Volume 1: RITECH},
year={2025},
pages={224-230},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0014276300004928},
isbn={978-989-758-784-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Innovations in Information and Engineering Technology - Volume 1: RITECH
TI - Improving Tsunami Prediction Accuracy Through Stacking Ensemble and Machine Learning Data Optimization
SN - 978-989-758-784-9
AU - Firmansyah H.
AU - Nurtanio I.
AU - Areni I.
AU - Purba J.
PY - 2025
SP - 224
EP - 230
DO - 10.5220/0014276300004928
PB - SciTePress