Workshop: Interactive data visualisation in Python 🐍

This repository contains the material for a workshop that I have run for the Centre de Recherches Mathématiques, the Centre for Applied Mathematics in Bioscience and Medicine, and the Quantitative Life Sciences Dept. at McGill University.

Instructor

Dr. Thomas Bury
Postdoctoral fellow
Department of Physiology, McGill University
Contact: thomas.bury@mcgill.ca

Summary

Modern scientific methods give rise to vast quantities of data. Creating effective visualisations is essential for both presentation and exploration of the data. This is no easy task when the data contains dozens of variables and millions of entries. Traditional visualisations are static, that is, what the user sees is what the user gets. Interactive visualisations allow the user to perform tasks such as varying parameters, honing in subsections of the data, and switching between different plot types, all without touching the code! This faciliates rapid exploration of the data. Moreover, these visulisations can be easily shared with collaborators who only require a web browser to open the visualisation.

This workshop will equip participants with the skills required to begin creating interactive visualisations in Python, using the Plotly library. The format will be interactive, with alternation between demonstrations by the instructor and participants working through their own Jupyter notebook (provided in advance). Participants will come away having made several of their own visualisations of either a large public dataset, or their own dataset if they would like to bring one. Check out this interactive visualisation of ECG data from Physionet for an example of what can be achieved with these tools.

Objectives

We will cover the following:

• Organisation of pandas DataFrames to facilitate rapid plotting.
• Creating basic interactive plots in Plotly (Bar, Line, Scatter, Histogram).
• Visualisation of multiple dimensions using multiple traces, grid plots and 3-dimensional graphs.
• Integration of sliders and drop-down boxes.
• (If time allows.) Combining interactive plots into a dashboard.

Setup instructions

1. Install the Anaconda distribution available here. This comes with Python 3, JupyterLab and the Python packages numpy and pandas, which we will require for the workshop.

2. Install Python packages for plotly and dash by entering the following commands into your Terminal (Mac) or Command Prompt (PC):

   conda install plotly
   conda install -c conda-forge dash
   conda install -c conda-forge -c plotly jupyter-dash
   

3. Install git if not already on your computer.

4. Make a Github account if you don't already have one.

5. Go to the workshop repository, and create a fork (button on top right). This creates your own version of the repository, in which you can make edits and complete the workshop notebooks.

6. Clone your repository to your computer. To do this, click the green button that says 'Code', and copy the link that appears in the box. Then go to your Terminal (or Command Prompt), navigate to a folder where you would like to store the repository and enter

   git clone paste-your-link-here
   

   where you paste your link as indicated. You should now see the directory workshop_datavis_python saved on your computer.

7. Open the Anaconda Navigator and launch JupyterLab. Navigate to where you saved the workshop repository and open test.ipynb. Check that you can run this notebook without any errors.

If you run into issues, please contact me at thomas.bury@mcgill.ca and I'll glady help out.