This tutorial is designed for data scientists, analysts, and researchers interested in advanced time series analysis and forecasting techniques. We'll explore cutting-edge methods for enhancing uncertainty quantification and prediction accuracy in time series data, using state-of-the-art tools like tsbootstrap and sktime.
Throughout this 4-hour session, we'll cover:
- Fundamentals of modern time series analysis and bootstrapping methods
- Advanced forecasting techniques and their practical applications across various domains
- Innovative approaches to uncertainty quantification and probabilistic forecasting
- Hands-on exercises using leading-edge tools to solve real-world time series challenges
By the end of this tutorial, you'll have gained insights into the latest developments in time series analysis and forecasting, and acquired practical skills to apply these advanced methods to your own data-driven projects.
In the tutorial, we will move through notebooks section by section.
You have different options how to run the tutorial notebooks:
- 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.
| Title | Description | |
|---|---|---|
| 1 | Overview of tsbootstrap and sktime | Feature overview of sktime and tsbootstrap |
| 2 | Forecasting with sktime | Introduction to forecasting with sktime. Basic usage, univariate/multivariate, endogenous/exogenous, probabilistic, hierarchical forecasts. |
| 3 | Time Series Bootstrapping with tsbootstrap | Introduction to bootstrapping, relation to uncertainty quantificaiton. Time series bootstrapping. |
| 4 | Probabilistic forecasts with sktime | Probabilistic forecasting vignettes. Backtest-evaluation of probabilistic forecasters, tuning. |
| 5 | Advanced probabilistic prediction | Deep-dive into prediction types, distributional predictions. Performance metrics revisited. Probabilistic reduction models, feature engineering. |
If you're interested in contributing to tsbootstrap or sktime, you can find out more how to get involved here and 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-scipy-2024
- Create a python virtual environment:
conda create -y -n sktime_tsbootstrap python=3.10 - Install required packages:
conda install -y -n sktime_tsbootstrap pip sktime tsbootstrap skpro seaborn jupyter pmdarima statsmodels - Activate your environment:
conda activate sktime_tsbootstrap - If using jupyter: make the environment available in jupyter:
python -m ipykernel install --user --name=sktime_tsbootstrap
- Create a python virtual environment:
python -m venv sktime_tsbootstrap - Activate your environment:
source sktime_tsbootstrap/bin/activatefor Linux- sktime_tsbootstrap/Scripts/activate` for Windows
- Install the requirements:
pip install -r requirements.txt - If using jupyter: make the environment available in jupyter:
python -m ipykernel install --user --name=sktime_tsbootstrap