Analysis of look back period for stock price prediction with RNN variants: A case study on banking sector of NEPSE
| Autor: | Arjun Singh Saud, Subarna Shakya |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Stock market prediction
business.industry Computer science Deep learning Future value 020206 networking & telecommunications 02 engineering and technology Profit (economics) Recurrent neural network Stock exchange 0202 electrical engineering electronic engineering information engineering Econometrics General Earth and Planetary Sciences 020201 artificial intelligence & image processing Stock market Artificial intelligence business Stock (geology) General Environmental Science |
| Zdroj: | Procedia Computer Science. 167:788-798 |
| ISSN: | 1877-0509 |
| DOI: | 10.1016/j.procs.2020.03.419 |
| Popis: | Stock market prediction is an attempt of determining the future value of a stock traded on a stock exchange. Stock market investors try to predict the stock’s future price to make trading decisions such that optimum profit can be earned. Deep learning models are found most successful in predicting stock prices. This paper has performed a novel analysis of the parameter look-back period used with recurrent neural networks and also compared stock price prediction performance of three deep learning models: Vanilla RNN, LSTM, and GRU for predicting stock prices of the two most popular and strongest commercial banks listed on Nepal Stock Exchange (NEPSE). From the experiments performed, it is found that GRU is most successful in stock price prediction. In addition, the research work has suggested suitable values of the look-back period that could be used with LSTM and GRU for better stock price prediction performance. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|