antoinecarme / pyaf

PyAF is an Open Source Python library for Automatic Time Series Forecasting built on top of popular pydata modules.
BSD 3-Clause "New" or "Revised" License
456 stars 73 forks source link

Investigate other multi-step forecasting strategies #84

Open antoinecarme opened 6 years ago

antoinecarme commented 6 years ago

PyAF uses an iterated one-step ahead forecasting, that is , the same model (signal transformation + signal decomposition) forecast is iterated one-step at a time.

Other forecasting strategies do exist :

  1. iterated one-step ahead forecasting (PyAF uses this one)
  2. direct -step ahead forecasting; and
  3. multiple input multiple output models.

see

  1. A review and comparison of strategies for multi-step ahead time series forecasting based on the NN5 forecasting competition By Souhaib Ben Taieb, Gianluca Bontempi, Amir Atiya and Antti Sorjamaa https://arxiv.org/pdf/1108.3259.pdf

  2. Long-term forecasting with machine learning models by Thomas Huijskens https://thuijskens.github.io/2016/08/03/time-series-forecasting/

antoinecarme commented 5 years ago

Strategy One. Recursive

image

antoinecarme commented 5 years ago

Strategy 2. Direct

image

antoinecarme commented 5 years ago

Strategy 3. MIMO / Joint

image

image