skedm

Scikit Empirical Dynamic Modeling

Scikit Empirical Dynamic Modeling (skedm) can be used as a way to forecast time series, spatio-temporal 2D or 3D arrays, and even discrete spatial arrangements. More importantly, skedm can provide insight into the underlying dynamics of a system, specifically whether a system is nonlinear and deterministic or whether it is dominated by noise.

For a quick explanation of this package, I suggest checking out the Quick Example section as well as the wikipedia article on nonlinear analysis . Additionally, Dr. Sugihara’s lab has produced some good summary videos of the topic:

1. Time Series and Dynamic Manifolds
2. Reconstructed Shadow Manifold

For a more complete background, I suggest checking out Nonlinear Analysis by Kantz as well as Practical implementation of nonlinear time series methods: The TISEAN package.

This software is useful both for forecasting and exploring underlying dynamical processes in a broad range of systems. The target audience is also wide-ranging as the software can be used to explore any dynamical system. Previous work using similar analyses has explored ecological systems, physical systems, and physiological applications.

Contents:

• Install
  • pip
  • Conda (Recommended)
  • Contribute, Report Issues, Support
• Quick Example
• Generate Data
  • 1 Dimensional
  • 2 Dimensional
• Embed
  • 1D Embedding
  • 2D Embedding
• Predict
• Score
  • Regression
  • Classification
  • Distinguishing Determinism
• Module Reference
  • utilities
• Performance
  • Near Neighbor
  • Train Size
  • Test Size
• Acknowledgements