Set up Kedro¶
Installation prerequisites¶
-
Python: Kedro works on macOS, Linux, and Windows and requires Python 3.9+. You should create a new virtual environment for each new Kedro project you work on to isolate its Python dependencies from those of other projects.
-
git: You must install
gitonto your machine if you do not already have it. Typegit -vinto your terminal window to confirm it is installed; it will return the version ofgitavailable or an error message. You can downloadgitfrom the official website.
Python version support policy¶
- The core Kedro Framework supports all Python versions that are actively maintained by the CPython core team. When a Python version reaches end of life, support for that version is dropped from Kedro. This is not considered a breaking change.
- The Kedro Datasets package follows the NEP 29 Python version support policy. This means that
kedro-datasetsgenerally drops Python version support beforekedro. This is becausekedro-datasetshas a lot of dependencies that follow NEP 29 and the more conservative version support approach of the Kedro Framework makes it hard to manage those dependencies properly.
Quickstart¶
The simplest way to get started with Kedro is to create a new project using our starters,
for example our Spaceflights starter.
To do that, we recommend using uv,
a fast, modern Python package and project manager.
First, use uvx to seamlessly run kedro new:
uvx kedro new --starter spaceflights-pandas --name spaceflights
The command will create a directory with the contents of your project. Navigate to it:
cd spaceflights
And finally, verify that everything works:
uv run kedro run --pipeline __default__
How to verify your Kedro installation¶
Note
For the sake of brevity, the rest of the Kedro documentation assumes
that you either run kedro through your project management tool,
for example executing uv run kedro,
or that the appropriate virtual environment is activated.
If you get "command not found" errors or other unexpected results,
double check that you are adding the prefixes required by your tooling
or follow standard troubleshooting steps first to verify that
the proper environment is activated.
To check that Kedro is installed:
kedro info
You should see an ASCII art graphic and the Kedro version number. For example:
If you do not see the graphic displayed, or have any issues with your installation, check out the searchable archive of Slack discussions, or post a new query on the Slack organisation.
Alternative methods¶
How to create a new Kedro project using kedro new¶
To create a new Kedro project using kedro new you will need Kedro installed first.
The best way to achieve this is to have a "global" installation of Kedro
that you can use only for this purpose.
There are several ways to do it:
uvx is an alias for uv tool, a uv feature that allows you to work with "tools"
or globally-installed Python packages. After installing uv,
you can use it to run kedro new as follows:
uvx kedro new
pipx is a tool that allows you to install and run Python applications and tools
in isolated environments. After installing it,
you can use it to run kedro new as follows:
pipx run kedro new
conda provides package, dependency, and environment management for any language.
Since it uses centralized environments, it is a good option for managing applications and tools.
After installing it,
you can use it to run kedro new as follows:
conda create --name kedro-environment python=3.13 kedro --yes
conda activate kedro-environment
kedro new
How to manually create a virtual environment for your Kedro project¶
There are two main types of virtual environments in Python:
native environments and conda environments.
The former are created with tools like venv, virtualenv and also uv,
and the latter are created with conda.
Tip
Read more about virtual environments for Python projects or watch an explainer video about them.
uv is also capable of creating virtual environments.
First, navigate to where your project is:
cd your-kedro-project
Next, create a new virtual environment inside the .venv subdirectory:
uv venv
Activate this virtual environment:
source .venv/bin/activate # macOS / Linux
.\.venv\Scripts\activate # Windows
To exit the environment:
deactivate
Modern versions of Python 3 include the venv module
in the standard library.
Otherwise you can install virtualenv.
First, navigate to where your project is:
cd your-kedro-project
Next, create a new virtual environment inside the .venv subdirectory:
python -m venv .venv
Activate this virtual environment:
source .venv/bin/activate # macOS / Linux
.\.venv\Scripts\activate # Windows
To exit the environment:
deactivate
Another popular option is to use Conda. After you install it, execute this from your terminal:
conda create --name kedro-environment python=3.13 -y
The example above uses Python 3.13, and creates a virtual environment called kedro-environment.
Notice that conda allows you to choose your Python version,
as well as the environment name.
The conda virtual environment is not dependent on your current working directory and can be activated from any directory:
conda activate kedro-environment
To confirm that a valid version of Python is installed in your virtual environment, type the following in your terminal:
python --version
To exit kedro-environment:
conda deactivate
How to manually install Kedro in an environment¶
Depending on your environment of choice, you can install Kedro using uv, other native tools (pip, uv, Poetry, PDM, and others), or conda.
You can use uv to manually install Kedro in a native virtual environment as follows:
uv pip install kedro
Alternatively, you can add it to your project so it gets automatically declared
in your project metadata (pyproject.toml) and a lock file gets generated:
uv add kedro
To install Kedro in a native Python environment from the Python Package Index (PyPI):
pip install kedro
You can use any modern Python packaging tool you want, such as Poetry, PDM, or Hatch:
poetry add kedro
You can also install Kedro using conda as follows:
conda install -c conda-forge kedro
How to manually install the project dependencies in an environment¶
Kedro projects have a pyproject.toml that specifies their dependencies
and enable sharable projects by ensuring consistency across Python packages and versions.
A requirements.txt file is also included for compatibility with legacy workflows.
This is a sample of the dependencies found in the Spaceflights Kedro project (you may find that the versions differ slightly depending on the version of Kedro):
# code quality packages
ipython~=8.10; python_version >= '3.8'
ruff==0.1.8
# notebook tooling
jupyter~=1.0
jupyterlab_server>=2.11.1
jupyterlab~=3.0
# Pytest + useful extensions
pytest-cov~=3.0
pytest-mock>=1.7.1, <2.0
pytest~=7.2
# Kedro dependencies and datasets to work with different data formats (including CSV, Excel, and Parquet)
kedro~=0.19.0
kedro-datasets[pandas-csvdataset, pandas-exceldataset, pandas-parquetdataset]>=3.0
kedro-telemetry>=0.3.1
kedro-viz~=6.0 # Visualise pipelines
# For modeling in the data science pipeline
scikit-learn~=1.0
You can use uv to sync the environment with the project dependencies:
uv sync
Alternatively, you can use the lower level interface to install the dependencies
directly from the requirements.txt file:
uv pip install -r requirements.txt
You can use pip or similar tools to install the dependencies
directly from the requirements.txt file:
pip install -r requirements.txt
Warning
The official Kedro starters don't ship with environment.yml files
that can be consumed by conda.
Use uv or other native tools instead.
Other setup steps¶
How to integrate Kedro in your IDE¶
Working in an IDE can be a great productivity boost.
For VS Code Users: Check out Set up Visual Studio Code and Kedro VS Code Extension.
For PyCharm Users: Checkout Set up PyCharm.
How to upgrade Kedro¶
Kedro follows the Semantic Versioning paradigm. Upgrades between bugfix or patch releases of the same minor series are expected to not require any special user action, otherwise it's considered a bug. Upgrades between minor releases of the 0.x series, or major releases after 1.0, can include backwards-incompatible changes that may require some user intervention to achieve a clean upgrade.
The best way to safely upgrade when backwards-incompatible changes are expected is to check our release notes. Follow the steps in the migration guide included for that specific release.
Apart from installing the latest version (see above),
when migrating an existing project to a newer Kedro version
make sure you also update the kedro_init_version:
- For projects generated with versions of Kedro > 0.17.0, you'll do this in the
pyproject.tomlfile from the project root directory. - If your project was generated with a version of Kedro <0.17.0, you will instead need to update the
ProjectContext, which is found insrc/<package_name>/run.py.
Once Kedro is installed, you can check your version as follows:
kedro --version
Summary¶
- Kedro can be used on Windows, macOS or Linux.
- Installation prerequisites include a Python 3.9+ and
git. - The simplest way to get started with Kedro is to use create a new project using our starters.
uvis a great choice for managing Kedro projects, but you can use other tools such aspip,Poetry,conda, and more.
If you encounter any problems as you set up Kedro, ask for help on Kedro's Slack organisation or review the searchable archive of Slack discussions.