Combine Facebook Prophet and LSTM with BPNN Forecasting financial markets: the Morgan Taiwan Index
| Autor: | Po-Chao Lan, Wan-Rung Lin, Yi-Hsien Wang, Hai-Yen Chang, Wen-Xiang Fang, Hsiao-Chen Chang |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
Index (economics)
Artificial neural network business.industry Computer science Deep learning 05 social sciences Financial market 02 engineering and technology Machine learning computer.software_genre Recurrent neural network Order (exchange) 0502 economics and business 0202 electrical engineering electronic engineering information engineering 050211 marketing 020201 artificial intelligence & image processing Artificial intelligence Time series Set (psychology) business computer |
| Zdroj: | ISPACS |
| DOI: | 10.1109/ispacs48206.2019.8986377 |
| Popis: | The recurrent neural network (RNN) used by many people in machine learning often faces the situation where the gradient disappears, in order to solve this problem. Modern scholars often use Long Short-Term Memory (LSTM) proposed in 1997 to predict time series samples. However, Facebook believes that most of the past time series models have missing adjustment parameters. Therefore, it developed a set of predictive tools Prophet for periodic parameters and trend parameters in 2017. In this paper, LSTM and Prophet are used to predict the trend of time series data, and the prediction trend is combined with the inverse neural network model (BPNN) for prediction. The empirical results show that this method can indeed achieve accurate forecasting trends and reduce errors. This research promises to contribute to this research literature in the future, thereby enhancing the ability of investors to target the long-term layout. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|