Probablistic Machine Learning - Summer 2023 - Project

We apply Gaussian Processes to predict daily climate data.

Quick Start

Google Colab

Discover our notebooks using Google Colab:

• Basics of Gaussian Processes
• Main Analysis

Local Setup

1. Start a Jupyter server by running docker run -p 8888:8888 -v $(pwd):/home/jovyan/work jupyter/scipy-notebook. Make sure you have docker up and running. Token-authentication is enabled, meaning you can access the application by opening http://127.0.0.1:8888/lab?token=<YOUR_TOKEN>. The URL, including your token, is displayed in your terminal.
2. Open src/GP_Main.ipynb and explore our data analysis 🔥

Developer Guide

Setup

Install all dependencies

pip install --user pipenv
pipenv install --dev

Extract the corresponding virtual environment and adapt the corresponding configuration of your IDE:

pipenv --py

You should be good to go! 🐥 We use the VS Code Jupyter Extension to run our notebooks.

Quality Guidelines

We use two main tools to assure code quality 😇

• black for formatting: pipenv run black src
• pylint for linting: pipenv run pylint src

Both is checked by a GitHub Actions pipeline (see .github/workflows/build.yml). If you work with Jupyter Notebooks, check out nbAQ.

Versioning

Since comparing pull requests can be quite tricky using Jupyter Notebooks, we use reviewNB to display diffs.