Skip to main content

Time Series Forecasting and Data Augmentation using Deep Generative Models

Project description

TAug :: Time Series Data Augmentation using Deep Generative Models

Note!!! The package is under development so be careful for using in production!

Features

  • Time Series Data Augmentation using Deep Generative Models
  • Visualizing the Latent Space of Generative Models
  • Time Series Forecasting using Deep Neural Networks

Installation

You can install the last stable version using pip

pip install taug

How to Use

Augmentation Guide

Create an augmenter

from taug.augmenters.vae import LSTMVAE
from taug.augmenters.vae import VAEAugmenter

# create a variational autoencoder
vae = LSTMVAE(series_len=100)
# use the created vae as an augmenter
augmenter = VAEAugmenter(vae)

The above code uses the default settings for the LSTM-VAE model. You can customize its architecture or use your own model for encoder and decoder. Note currently we only support Keras models.

Train the augmenter

augmenter.fit(data, epochs=64)

Generate new time series!

A few strategy for sampling have been implemented.

Sampling from whol
augmenter.sample(n=1000)

Forecasting Guide

[todo] Forecasting guide will be here!

Supported Augmenters

Supported models for augmentation currently are as follows:

Model Type Supported Time Series Description
LSTMVAE Variational Autoencoder Univariate, fixed length A Variational Autoencoder with stacked LSTM layers for encoder and decoder

Supported Forecasters

Supported models for time series forecasting are as follows:

Contributors

The list of the current contributors:

  • Sasan Barak
  • Amirabbas Asadi

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

taug-0.1.0.tar.gz (8.0 kB view hashes)

Uploaded Source

Built Distribution

taug-0.1.0-py3-none-any.whl (8.4 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page