Прогнозування часових рядів методом LSTM
| dc.contributor.advisor | Щестюк, Наталія | |
| dc.contributor.author | Поляков, Михайло | |
| dc.date.accessioned | 2022-01-17T21:27:27Z | |
| dc.date.available | 2022-01-17T21:27:27Z | |
| dc.date.issued | 2021 | |
| dc.description.abstract | Time-series forecasting is a complex problem. Forecasting includes using models to fit historical information and using it to predict future data points. There are multiple methods of forecasting time-series data, including autoregressive models, like ARIMA, using gradient boosting like XGBoost, and more recently, using neural networks. The different techniques might be better for different types of data, and most of the time, only comparison between models can show the best model. We would try creating an LSTM based neural network with a multi-step multiple output strategy in this work. The completed model is tested on two different datasets in nature, power usage, and stock market prices. | uk_UA |
| dc.identifier.uri | https://ekmair.ukma.edu.ua/handle/123456789/22286 | |
| dc.language.iso | en | uk_UA |
| dc.status | first published | uk_UA |
| dc.subject | Time-series forecasting strategies | uk_UA |
| dc.subject | LSTM | uk_UA |
| dc.subject | магістерська робота | uk_UA |
| dc.title | Прогнозування часових рядів методом LSTM | uk_UA |
| dc.title.alternative | Time-series forecasting using LSTM | uk_UA |
| dc.type | Other | uk_UA |