PROBABILISTIC NEURAL NETWORK FOR ANOMALY DETECTION IN MOUNT MERAPI'S SEISMIC ACTIVITY: PERFORMANCE EVALUATION AND INSIGHTS

Paramitha Nerisafitra (Teknik Informatika, Universitas Negeri Surabaya)
Tri Deviasari Wulan (Sistem Informasi, Universitas Nahdlatul Ulama Surabaya)
Ricky Eka Putra (Teknik Informatika, Universitas Negeri Surabaya)
I Made Suartana (Teknik Informatika, Universitas Negeri Surabaya)

ICONESTH
2024: The 2nd ICONESTH

This research aims to evaluate the performance of the Probabilistic Neural Network (PNN) method for detecting seismic anomalies in the monitoring data of Mount Merapi. The study utilized a dataset comprising 368 records, representing both normal activity and increased seismic activity. The dataset was divided into 70% for training and 30% for testing. During the training phase, the PNN model achieved an accuracy of 87%, indicating its capability to identify patterns in the seismic data effectively. However, the testing phase, conducted to validate the model’s generalization ability, yielded an accuracy of 64%. These results suggest that while the PNN method demonstrates promise in detecting seismic anomalies, its performance requires further improvement to enhance reliability in operational volcanic monitoring systems. Keywords: probabilistic neural network, seismic, performance