sebastian-ruiz/deep-learning-simulating-time-series-futures

Code for: Application of Deep Learning for Simulating Time-Series of Oil Futures

This repository contains the code for the master's thesis: Application of Deep Learning for Simulating Time-Series of Oil Futures (link to follow), by Sebastian Ruiz, for the program mathematics, stochastics track, at the University of Amsterdam.

Note: The time-series data of oil futures is purposely missing from this project as the data is not publicly available.

Getting Started

• Set PATH_ROOT correctly in .env (make a copy of .env.sample).

If importing data using the Calibration Library:

1. Add CalibrationLib to project interpreter path. In PyCharm: Settings -> Project interpreter -> Cog -> Show All -> Select your interpreter -> Click funny folder button -> Add -> Select CalibrationLib
2. Set writableDir correctly in data_importer/config.yaml
3. Run data_importer/importer_run.py

Training the models:

1. Set the training and test data in (root) config.yaml.
2. To train all the autoencoder models run autoencoders/autoencoder_run_all.py.
3. To train all the GAN models run gans/gan_run_all.py.

For each model the parameters can be set in the dictionary ae_params and gan_params for autoencoder parameters and GAN parameters respectively..

Helper files

• plotting.py: Provides function to plot curves. Used to compare autoencoder input and output and show GAN simulations.
• preprocess_data.py: Load data from pickle file and does the data pre-processing. It splits the data into training and test sets, and it applies normalisation, standardisation or log-returns.

Autoencoders

The code for the models is based on examples found in keras-autoencoders .

• Standard Autoencoder: Standard autoencoder with encoder and decoder.
• Variational Autoencoder (VAE): The model from Auto-Encoding Variational Bayes.
• Adversarial Autoencoder (AAE): The model from Adversarial Autoencoders.

GANs

Popular conditional GAN models. The code is based on examples found in Keras-GAN.

• Standard GAN: The generative adversarial network model, from Generative adversarial networks.
• Wasserstein GAN: The model from Wasserstein GAN.
• GAN CONV: GAN using convolutional layers.

LSTMs

The LSTM Model is based on example code from Keras seq-2-seq Signal Prediction.

GAIN

The GAIN model, GAIN: Missing Data Imputation using Generative Adversarial Nets, is based on the Tensorflow example code from GAIN.

Detect Anomalies

The anomaly detection model uses the autoencoder model.

Classical Models

The classical models include the Andersen Markov Model and some autoregressive models.