An implementation of Gaussian processes in Python. A model for supervised learning, GP, as well as a model for unsupervised learning, GPLVM, are provided. Multiple kernels are implemented, along with gradients to optimise hyperparameters.
Running the code requires Python 3.6+ and the following packages: NumPy, Matplotlib, SciPy and scikit-learn. To install them with pip, run
$ pip3 install -e .
This will install the required packages, and this project in editable mode, making the examples runnable from terminal.
Alternatively, install with pipenv using the provided Pipfile:
$ pipenv sync
Two examples are included, plot_gp.py and plot_gplvm.py. These can be run from terminal, and can be modified to try some other data and/or kernel. Run the examples from the project root, as below:
$ python3 gp/examples.plot_gp.py
The default setting is interactive, when you add new data points by clicking in the plot. The model will update to account for all new observations.
![@dependabot[bot]](https://avatars.githubusercontent.com/in/29110?s=64&v=4){kind=small}