Binder Tutorial

Repository/Working directory

First, a directory for the project should be created and turned into a Git repository, or a git repository could be created on GitHub and cloned to the local machine.

git clone https://github.com/sVujke/{repo_name}.git

Setting up the environment

Now, an environment should be created, this time we will do it with Anaconda.

conda create -n binder_env python=3.6

This will create an environment arbitrarilly named as binder_env, with a python 3.6 interpreter. Before installing the packages, we need to activate the environment!

activate binder_env (Windows)
source activate binder_env (macOS or Linux)

In this example, two libraries have been added:

conda install pandas
conda install tensorflow

NOTE! In order for the jupyter notebook to recognize the environment, we need to install ipykernel

conda install ipykernel

We need to add pur environment as a kernel, which will enable jupyter notebook to recognize our environment and it's interpreter.

 python -m ipykernel install --user --name binder_env --display-name "Py - binder"

After this, we will be able to import packages into our notebooks!

Exporting the environment

As we need to keep track of our dependencies, we need to export a list of packages from our environment:

conda env export > environment.yml

After this, we should commit and push our changes.

Using Binder

All you need is a link to your GitHub repo, and you need to add it to the specified field on https://notebooks.gesis.org/binder/ or https://mybinder.org/ alternatively.

I would recommend removing python versions for the libraries as it currently tends to create problems.

Launch the notebook!

Further reads

Using Binder