The goal of this notebook is to demonstrate the dynamic mode decomposition by forecasting time series data from the M5 forecasting competition. The dynamic mode decomposition is a dimensionality reduction algorithm developed by Peter Schmid in 2008.
git clone https://github.com/KennethJAllen/dynamic-mode-decomposition
cd dynamic-mode-decomposition
- Install Poetry if not already installed.
- Run the following command in the project directory:
poetry install
poetry shell
You can now run the project's scripts within the poetry shell.
The dynamic mode decomposition, originally developed for fluid dynamics simulation, was designed to extract features from high dimensional data.
Given a collection of
The M5 competition was a times series forecasting competition held in 2020, with the objective of advancing the theory and practice of time series forecasting methods. The M5 data, made available by Walmart, consists of the unit sales of