Institutional Repository of Vilnius University
Dirbtiniai neuroniniai tinklai laiko eilučių prognozavimui meteorologijoje /
Abstract
dc:descriptionMeteorology is the study of the Earth’s atmosphere and its phenomena, with the aim of understanding weather patterns, climate change and providing a basis for making predictions. In this field, forecasts are usually made using numerical weather prediction, which requires large amounts of data and processing power. This study evaluated the accuracy of neural networks for this task, aiming to reduce the resources needed for forecasting. Three neural network architectures were chosen for the study: Multilayer Perceptron, Gated Recurrent Unit model, and Neural Basis Expansion Analysis for Time Series. The accuracy of these models was evaluated using the Mean Absolute Error and Mean Squared Error metrics, and the results were compared with the predictions of statistical Seasonal Naive and Autoregressive Integrated Moving Average models. The study found that neural networks are effective in forecasting long and short term air temperatures in Lithuanian cities, outperforming both statistical methods in short-term forecasts but occasionally being outperformed by the statistical ARIMA model in long-term forecasting.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Slivinskas, Povilas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD210648032&prefLang=en_US