This tutorial is about sktime - a unified framework for machine learning with time series. sktime contains algorithms and tools for building, applying, evaluating modular pipelines and composites for a variety of time series learning tasks, including forecasting, classification, regression.
sktime is easily extensible by anyone, and interoperable with the python data science stack.
This is an introductory sktime half-day tutorial with:
- a general introduction to
sktime - forecasting with
sktime- uni/multivariate, hierarchical/global, probabilistic - feature extraction, transformation pipelines, parameter tuning
- advanced pipeline including autoML and graphical pipelines
- evaluation and benchmarking in sktime.
In the tutorial, we will move through notebooks section by section.
You have different options how to run the tutorial notebooks:
- Run the notebooks in the cloud on Binder - for this you don't have to install anything!
- Run the notebooks on your machine. Clone this repository, get conda, install the required packages (
sktime,seaborn,jupyter) in an environment, and open the notebooks with that environment. For detail instructions, see below. For troubleshooting, see sktime's more detailed installation instructions. - or, use python venv, and/or an editable install of this repo as a package. Instructions below.
Please let us know on the sktime discord if you have any issues during the conference, or join to ask for help anytime.
This workshop presents sktime - a unified framework for machine learning with time series. sktime covers multiple time series learning problems, including time series transformation, classification and forecasting, among others.sktime allows you to easily apply an algorithm for one task to solve another (e.g. a scikit-learn regressor to solve a forecasting problem). In the tutorial, you will learn about how you can identify these problems, what their key differences are and how they are related.
This tutorial is an introduction to advanced forecasting techniques using sktime and its objectives encompass:
- Learn how to create forecasting pipelines that integrate forecasters and feature extraction methods.
- Explore more sophisticated and advanced forecasting scenarios and realise them using graphical pipelines and auto-ML.
- Learn how to evaluate the performance of forecasters.
- Learn how benchmarks can be created, similar to M4/M5 competitions.
Additionally, this tutorial provides an introduction to reproducibility tools, such as auditably storing model blueprints and fitted models together with a methodological primer.
sktime not just a package, but also an active community which aims to be welcoming to new joiners.
We invite anyone to get involved as a developer, user, supporter (or any combination of these).
-
Pydata London 2022 - How to implement your own estimator in sktime
-
Pydata London 2023 - Time Series Classification, Regression, Distances & Kernels
If you're interested in contributing to sktime, you can find out more how to get involved here.
Any contributions are welcome, not just code!
To run the notebooks locally, you will need:
- a local repository clone
- a python environment with required packages installed
To clone the repository locally:
git clone https://github.com/sktime/sktime-workshop-pydata-prague-2023
- Create a python virtual environment:
conda create -y -n pycon_prague python=3.9 - Install required packages:
conda install -y -n pycon_prague pip sktime seaborn jupyter pmdarima statsmodels dtw-python - Activate your environment:
conda activate pycon_prague - If using jupyter: make the environment available in jupyter:
python -m ipykernel install --user --name=pycon_prague
- Create a python virtual environment:
python -m venv pycon_prague - Activate your environment:
source pycon_prague/bin/activatefor Linuxpycon_prague/Scripts/activatefor Windows
- Install the requirements:
pip install -r requirements - If using jupyter: make the environment available in jupyter:
python -m ipykernel install --user --name=pycon_prague