When models are deceptively bad.

Posted on Tue 17 Sep 2024 in data-analysis

I was once tasked with fitting some machine learning models in R to a subset of the well-known abalone dataset, then recommending the best one. The aim was to use non-invasive measurements (like dimensions and weight) to classify their sex.

Ultimately, I recommended none of the models. Their overall accuracy …


Continue reading

How I keep my R projects reproducible.

Posted on Fri 02 Aug 2024 in computing

Unless you are very lucky, replacing your version of R with the most recent one, or continually upgrading your packages, will eventually break your projects. I am apparently not very lucky, because this has happened to me - my university upgraded its packages while I was mid-assignment, and my code suddenly …


Continue reading

Auto-updated visualisations of COVID-19 cases by geographical area.

Posted on Wed 20 Dec 2023 in data-analysis

In 2020, during the first significant COVID-19 wave to have occurred in Victoria, Australia, I created an automatically updating interactive of daily case numbers, by LGA (local government area).

In addition, I created videos showing showing how case numbers evolved across Greater Melbourne and Victoria.

These are published here.

The …


Continue reading