archercym / ceemdan-lstm

Paper-Reproduce: (ESWA) Forecasting the realized volatility of stock price index: A hybrid model integrating CEEMDAN and LSTM

CEEMDAN-LSTM

Paper-Reproduce: (ESWA) Forecasting the realized volatility of stock price index: A hybrid model integrating CEEMDAN and LSTM

---

File Information

1. "P888.csv" is raw data, including <datetime, trading_date, low, volume, high, close, open, total_turnover>. You can change it to any stock data.
2. "Forecasting the realized volatility of stock price index A hybrid model integrating CEEMDAN and LSTM.pdf" is the paper i reproduce.
3. "CEEMDAN-LSTM.ipynb" is all the code.

Methods Description

1. Clean data, generate RVs as target and make some statistic analysis (skewness, excess kurtosis, J-B, Q(10))
2. CEEMDAN decomposition and visualization. (PyEMD)
3. Build LSTM model on raw-data and decomposed data. (TensorFlow)
4. Make comparison with SVR, AR, HAR.

---

Tips

If you have any problem, you can send an email [email protected] or make an issue directly. We can discuss together.

ceemdan-lstm's Issues

LSTM模型预测值为0

版本咨询

您好，可以问一下代码中的python和tf版本吗，我在使用3.7python和2.1tfGPU时遇到了一些问题