Geographic Data Science - Lecture II

Modern Computational Environments

Dani Arribas-Bel

How does Science “get done”?

  • Reproducibility in Science
  • Modern scientific tools
  • JupyterLab demo

Reproducibility in Science

In the old days…

Reproducibility

  • Ability to reproduce scientific procedures (e.g. experiments, results)
  • Key to the scientific endeavour
  • Embedded in early work

But…

url

url

url

Modern Scientific Tools

Reproducible Science

  • Transparent (computational) processes
  • Enough detail to reproduce the entire analysis
  • Efficient model of reusability (D.R.Y.)

Building Blocks

  • Computational Notebooks
  • Open-Source Packages
  • Open Platforms

Computational Notebooks

url

Computational Notebooks

One-file documents:

  • (Executable) code
  • Output
  • Narrative text

Open-Source Packages

D.R.Y. (Don’t Repeat Yourself)

  • Encapsulate reusable functionality
  • Easy access + more reliable (if package is good!)
  • Code available (free as in beer… but also as in speech!)

Platforms

Hardware and low-level software (OS) that supports computations

Change of models:

  • Desktop Vs cloud
  • Integrated Vs distributed
  • Native installation Vs virtualisation/containerisation

Examples

Ligo gravitational waves

2019

JupyterLab (live) demo

Creative Commons License
Geographic Data Science’19 by Dani Arribas-Bel is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.