Skip to content

Time series prediction using dilated causal convolutional neural nets (temporal CNN)

License

Notifications You must be signed in to change notification settings

kristpapadopoulos/seriesnet

Repository files navigation

SeriesNet

Time series prediction using dilated causal convolutional neural nets

In this paper, state of the art deep learning techniques for time series forecasting were surveyed and a dilated causal convolutional neural network was developed (i.e. SeriesNet) based on the WaveNet architecture to forecast time series. It was found that SeriesNet without data preprocessing and ensemble methods achieved comparable results to top performing deep learning models on the international Computational Intelligence in Forecasting (CIF2016) competition dataset. This work extends previous work on using convolutional neural networks for time series forecasting.

Example of time series predictions with SeriesNet

  • First plot: log + normal noise - prediction for 20 steps in green
  • Second plot: log + sine - prediction for 20 steps in orange

log_noise log_sine

Requirements

  • tested with tensorflow 2.6.5
  • install requirements.txt
  • train.py contains data preparation, model training and prediction functions
  • toy example given in train.py to generate predictions and plot results