bgithub1 / da_rnn

RNN based on Chandler Zuo's implementation of the paper: A Dual-Stage Attention-Based Recurrent Neural Network for Time Series Prediction

da_rnn_from_csv (see below if the ipynb notebooks do not render in github)

(Only tested with python 3.5 and above)

---

This version of da_rnn is derived from Chandler Zuo's implementation of the paper:
A Dual-Stage Attention-Based Recurrent Neural Network for Time Series Prediction
which he details in his blog post
A PyTorch Example to Use RNN for Financial Prediction

My implementation does the following:

• You can replace the input data with other time series csvs.
  Change the input csv file by changing the first 2 lines of the code block in section 4.01.

        if __name__=='__main__':
           fname_no_ext = 'nasdaq100_padding'
           rows_to_use = 5000
           ...
  

• You can easily save the ipynb notebook as a python py file, that can be imported into other ipynb notebooks or other python projects.
  Use the bash command:
  jupyter nbconvert da_rnn_from_csv.ipynb --to python

• You can pickle the da_rnn class:
  pickle.dump(m,open(f'{fname_no_ext}.pt','wb'))
  Then reload it and use it to make more predictions:
  m_from_pickle = pickle.load(open(f'{fname_no_ext}.pt','wb'))

---

Use of da_rnn_from_csv.ipynb:

---

To use the da_rnn_from_csv.ipynb notebook, run all of the notebook's cells.
In section 4.0, you can change the data from nasdaq100 component data to uso 1-minute time bars.

The cell with the if __name__=='__main___ code (in section 4.0) will launch:

• Instantiating the model,
• Training of the model,
• Saving of the model,
• Reloading the model,
• Rerunning predictions using the saved model

---

Data:

The folder data contains several csv files:

• nasdaq100_padding.csv - original csv file used by Chandler Zuo of NDX components
• uso_full.csv - one minute bar data for the commodity ETF USO
• uso_201812.csv - one minute bar data just for December 2018

You can train the model using any of these csv files, or any other csv whose columns conform to those examples. Other time series csv files should only contain numeric columns. One of those numeric columns should be used as the label/y-value for the x-values of the preceeding row.

---

Structure of the csv/DataFrame:

The data in the csv/DataFrame can contain rows like:

1. the returns data for the components of an index like NDX (as in Chandler Zuo's version) as in the csv file ./data/nasdaq100_padding.csv, or
2. intra-day bar data with columns like year, month, day, hour, minute open, high, low, close, as in the csv file ./data/uso_full.csv

---

The main() method in section 3.0

The actual model training takes place in the main() method in the cell below heading 3.01, with the code:

    logger = setup_log()
    m = da_rnn(df_partial, logger = logger, parallel = False, learning_rate = .001)
    m.execute_training(n_epochs=100)

---

Using pdb

To use the pdb debuger, add pdb.set_trace() statements to the code.
See this cheatsheet for a quick reference to pdb commands

If you cannot view the ipynb notebooks in this project, see:

1. https://nbviewer.jupyter.org/github/bgithub1/da_rnn/blob/master/da_rnn_from_csv.ipynb
2. https://nbviewer.jupyter.org/github/bgithub1/da_rnn/blob/master/sine_curve_test.ipynb