/GAN_Time_Series

A model to generate time series data with the purpose of augmenting a dataset of various time series.

Primary LanguagePythonOtherNOASSERTION

GAN_Time_Series

The model is a Conditional Generative Adversarial Network for time series with not regular time intervals.

The model is created to generate a new time series given a training set of them.

Why generating data?

The main idea is to use this model to augment the unbalanced dataset of time series, in order to increase the precision of a classifier.

HOW TO USE THE MODEL

  • Requirements:

  • python 3

  • tensorflow, numpy

  • Download the repository

  • python3 main.py N M file_in file_times file_out

  • N = training set size

  • M = time series length

  • file_in = file input path

  • file_times = file with time stamps of the new time series

  • file_out = file output path

If you are interested in this work: https://arxiv.org/abs/1811.08295