The Level of Accuracy of Machine Learning Measurements in Predicting Earthquakes on Lombok Island

• Rizky Munandar

  Yogyakarta State University

  Author

Introduction: Lombok Island is one of the regions with the highest seismicity levels in Indonesia due to its geographical location trapped between two active earthquake sources: the subduction zone of the Indo-Australian plate with the Eurasian plate to the south and the Flores Back Arc Thrust Fault to the north. Predictions of earthquakes are still very minimal; the 2018 Lombok earthquake is one of the reasons why earthquake detectors need to be developed.

Objective: This study aims to analyze the use of machine learning in measuring the accuracy of earthquake predictions on the island of Lombok.

Methods: The method used in this study is comparative entrepreneurship which utilizes secondary data from the earthquake catalog during the January-October 2018 period to be analyzed using 3 machine larning algorithms, namely Naive Bayes, Artificial Neural Network (JST) and KNN.

Results: The results showed that the accuracy value using Naive Bayes was 0.6 and the accuracy using JST and KNN was 0.5. However, this is different from the results of the evaluation of the three algorithms where Naive Bayes still has a value of 0.6 but JST and KNN become 0.4.

Conclusion: In conclusion, the accuracy of machine learning measurements from the three algorithms shows that Naive Bayes has a high accuracy value, but this result may change if other algorithms are used.

• pdf

• 2026-01-02 (2)
• 2025-12-31 (1)