Liam-F / rnn_predictions

Building RNNs that generate sequences based on input data applied to time-series and text generation

Using RNNs: time series prediction and text generation

In this project I'll build Recurrent Neural Networks (RNNs) that can generate sequences based on input data - with a focus on two applications: With the first I'll use real market data in order to predict future Apple stock prices using a simple RNN model implemented in Keras. The second one will be trained on Sir Arthur Conan Doyle's classic novel Sherlock Holmes and will generate wacky sentences based on it that may - or may not - become the next great Sherlock Holmes novel =) This project is part of the Artificial Intelligence Nanodegree program, from Udacity. You can check my report here.

Install

This project requires Python 3 and the following Python libraries installed:

• Unidecode
• NumPy
• SciPy
• Scikit-learn
• Pandas
• Matplotlib
• Jupyter
• Tensorflow
• Keras

Run

In a terminal or command window, navigate to the top-level project directory aind2/ and run one of the following command:

    $ jupyter notebook RNN_project.ipynb

This will open the Jupyter Notebook software and notebook in your browser.

Main References

1. BLA, B. B. et al. Bla Bla. Bla Journal, 2017.

License

The contents of this repository are covered under the MIT License.