Article Per Year (5 Year)
p-Index From 2021 - 2026
0.817
P-Index
This Author published in this journals
All Journal JOIV : International Journal on Informatics Visualization JOURNAL OF APPLIED INFORMATICS AND COMPUTING Data Science: Journal of Computing and Applied Informatics
Saputri, Prilyandari Dina
Unknown Affiliation
Computer Science & IT

Published : 3 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search
Journal : JOIV : International Journal on Informatics Visualization

 1 
Prediction Intervals for Extreme Rainfall in Indonesia using Monotone Composite Quantile Regression Neural Networks Saputri, Prilyandari Dina; Azwarini, Rahmania; Adipradana, Dimaz Wisnu
JOIV : International Journal on Informatics Visualization Vol 9, No 6 (2025)
Publisher : Society of Visual Informatics

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62527/joiv.9.6.3186

Abstract

Rainfall data may contain nonlinear, complex, and extreme characteristics. Weather monitoring can be performed by predicting rainfall as the cause of flooding and providing early warnings to ensure smooth evacuation. Classical methods, such as ARIMA, are unable to capture rainfall data patterns. A standard method for forecasting complex datasets is the use of neural networks. The neural network method failed to produce a prediction interval due to the limitation of the standard error calculation. The use of the Monotone Composite Quantile Regression Neural Network (MCQRNN) enables the accommodation of complex patterns and the production of interval predictions through its quantiles. The crossing problems in the quantile estimation were also resolved. In this study, we utilized four rainfall datasets from different locations: Central Java, West Java, South Sumatra, and North Sumatra. The lower and upper bounds were compiled from 2.5% and 97.5%, respectively. The point forecasts are constructed from the 50% quantile. Furthermore, the point forecast and interval prediction were compared to the standard classical forecasting method, i.e., ARIMA. The results demonstrated that the MCQRNN model outperforms the ARIMA model in terms of point forecasting. As the forecasting period is extended, the interval prediction of MCQRNN tends to become more consistent, whereas the width prediction of the ARIMA model becomes broader. Hence, the MCQRNN interval predictions are also suitable for long-term forecasting. Further research was required to evaluate the performance of prediction intervals.
Co-Authors Adipradana, Dimaz Wisnu Alfajriyah, Aimmatul Ummah Angrenani, Arin Berliana Azwarini, Rahmania Fitriana, Ika Nur Laily Hakiki, M. Taufik Imam Safawi Ahmad, Imam Safawi Majid, Muhammad Althaf Mohamad Atok Muslichah Erma Widiana Siswono, Galuh Oktavia Soehardjoepri, Soehardjoepri Ulil Azmi Widyaningrum, Mahmudah Enny Yulius Hari