dfm / gp

A tutorial about Gaussian process regression

This repository contains an interactive IPython worksheet (worksheet.ipynb) designed to introduce you to Gaussian Process models. Only very minimal experience with Python should be necessary to get something out of this.

Some of this worksheet was originally prepared for a lab section at the Penn State Astrostats summer school in 2014 and it has been updated and adapted several times since then.

Remember: the best reference for anything related to Gaussian Processes is Rasmussen & Williams.

Prerequisites

You'll need the standard scientific Python stack (numpy, scipy, and matplotlib), a recent (3+) version of IPython/Jupyter (including the notebook), and emcee installed. If you don't already have a working Python installation (and maybe even if you do), I recommend using the Anaconda distribution and then running pip install emcee.

Usage

After you have your Python environment set up, download the code from this repository by running:

git clone https://github.com/dfm/gp.git

or by clicking here.

Then, navigate into the gp directory and run

cp worksheet.ipynb worksheet_in_progress.ipynb
jupyter notebook

This might open a web browser with the correct URL, but if not, you can copy and paste the URL that it prints to the terminal into your browser. Click on worksheet_in_progress.ipynb to get started.

License

This repository and the worksheet are copyright 2015-2017 Dan Foreman-Mackey and they are made available under the terms of the MIT license.