Skip to content

Latest commit

 

History

History

timegan

See the colab developed by @firmai for an interactive experience.


2019 NeurIPS Submission Title: Time-series Generative Adversarial Networks Authors: Jinsung Yoon, Daniel Jarrett, Mihaela van der Schaar

Last Updated Date: May 29th 2019 Code Author: Jinsung Yoon (jsyoon0823@gmail.com)

  1. Datasets (in Data Folder) (1) Google dataset
  • GOOGLE_BIG.csv (2) Sine dataset
  • Generated from data_loading.py
  1. Codes (1) data_loading.py
  • Transform raw time-series data to preprocessed time-series data (Googld data)
  • Generate Sine data

(2) Metrics Folder (a) visualization_metrics.py

  • PCA and t-SNE analysis between Original data and Synthetic data (b) discriminative_score_metrics.py
  • Use Post-hoc RNN to classify Original data and Synthetic data (c) predictive_score_metrics.py
  • Use Post-hoc RNN to predict one-step ahead (last feature)

(2) tgan.py

  • Use original time-series data as training set to generater synthetic time-series data

(3) main.py

  • Replicate the performances of Table 2 and Figure 3 in the paper
  • Report discriminative and predictive scores for each dataset and t-SNE and PCA analysis
  1. How to use? (1) In order to replicate the results in the paper
  • Run main.py (with selecting the dataset)

(2) In order to achieve time-series synthetic dataset (Main objective of the paper)

  • Run tgan.py
  • Input original time-series data to achieve corresponding synthetic time-series data