The constituencies of Israel
This project tries to simulate what the latest Israeli election would have been like under a single-member electorate system. If we use a “first-past-the-pos...
Blog
2024
Notes on Notes on the Synthesis of Form
Notes on the Synthesis of Form is a book written by architect and design theorist Christopher Alexander and originally published in 1964. It has become highl...
The Green Chain Walk
For many years now I have been reading Craig Mod’s writings about his many long, mostly solo, walks throughout Japan, where he has lived for many years. I fi...
2023
The relationship between technical debt and technical documentation
Technical debt is a phrase that gets used a lot when building software and software companies. It refers to the process where deliberate decisions are made t...
Learning two things at once - using Rust and ECS to build a cricket simulator
Here is a strategy for time-poor working parents who are trying to learn new things: try to learn two new things at once. That’s what I did recently when try...
This is your brain on ChatGPT
Imagine the not too distant future. All menial tasks have been outsourced from our brains to generative AI models like the one that powers ChatGPT. Always th...
Building iOS apps with React Native
Until quite recently I hadn’t got around to building a mobile app, but I had a vague idea about the “native” vs “web” debate. In iOS, “native” means the appl...
2022
Conquering the last piece of the stack with React
I first started programming when I was 16 (so 20 years ago). I have been writing mostly Python for a living, alongside SQL and a bunch of other backend and d...
The ubiquity of linear programming
NP-hard problems are the most interesting and important class of computational problems. In my experience (and many others’ I imagine), most hard optimizatio...
2021
Fifty years of Australian elections in an alternative electoral reality
Australia has a very unique electoral system, with geographic “districts” (almost universally referred to as electorates) modelled on the British Westminster...
The Setup
In this post I answer the interview questions from The Setup - again. This is an updated version of this post which I wrote in 2019.
The wonders of org-mode
I wrote last year about how I finally got around to building my own system of organising my notes, thoughts, writings and to-do lists based on emacs and org ...
2020
Multilevel regression with poststratification (MRP)
Following on from a post earlier in the year about multilevel models, I thought I’d follow that post up with a discussion about how multilevel models are com...
Deriving Kepler's laws from Newton's laws
I’ve always been interested in physics but somehow, despite my years of studying mathematics, I’d never really properly studied physics outside of a couple o...
Encrypting files in emacs org-mode
People who use org-mode in emacs for sensitive things like keeping a private journal, or recording their unprocessed thoughts on all manner of subjects, have...
Multilevel regression
This post is about what people usually mean when they use the term “heirarchical regression”, which is one of those confusing terms in statistics that can ha...
Getting started with emacs
I’ve been looking for a system of organising my notes, which represent a sort of “second brain” for my thoughts, for many years. I have thousands of text fil...
2019
The 2019 UK election under AV
On December 12 this year, there were parliamentary elections in the United Kingdom for the third time in five years. The country is in the middle of an unpre...
The Setup
In this post I answer the interview questions from The Setup. This is an updated version of this post which I wrote in 2016, which is itself an updated versi...
Regression
At work recently I gave a 90-minute colloquium talk on regression. We have a monthly series of talks where researchers and data scientists discuss fundamenta...
The curse of dimensionality and the weak law of large numbers
I was recently re-reading some fantastic course notes by the peerless Terence Tao on basic statistics from a fairly high mathematical standpoint as I wanted ...
2018
Deep learning and the universal approximation theorem
Deep learning, and in particular deep neural networks, is one of the most popular and powerful tools of modern data science. Using analogies from neuroscien...
2017
The 2016 Australian election - New Zealand rules
New Zealand has a significantly different electoral system to Australia, and just had an interesting election with a fairly unexpected result. I thought it m...
Fixing the Trendy scraper
My Google trends scraper is the most popular post on this site, and I’ve been getting questions about it for the last year or so, ever since Google changed t...
The Bias-Variance tradeoff
The Bias-Variance tradeoff is an extremely important concept in statistical modelling which is often misinterpreted or poorly understood. In this post I’ll g...
2016
Installing MediaWiki on a DigitalOcean droplet (including Wikipedia navbars)
This post will explain from scratch how to install MediaWiki on a DigitalOcean droplet, including how to set up Wikipedia-style navigation bars. It’s somethi...
The Setup
In this post I answer the interview questions from The Setup. This is an updated version of this post which I first wrote here in April 2015.
Call your mum
Since moving overseas away from my family and friends nearly two years ago, one problem I have had is keeping in touch with everyone. A combination of distan...
Australian federal election post-mortem
The below graph of the ALP and Coalition two-party preferred vote from the 2013 election until the eve of the 2016 election contains a massive amount of info...
2015
Lifetime flight simulator
I spent a few hours today writing a basic first draft of a data-driven text-based “game” which demonstrates just how safe air travel is. The code for the gam...
Forecasting Australian elections - the tricky seats
In this post I want to look at the most difficult seats from the 2013 election for my basic model: those eleven seats whose final two candidates in the runof...
Marginal seat polls and the strong transition model
As I’ve recently (and finally) become gainfully employed, I haven’t been able to devote a huge amount of time to my model for forecasting Australian election...
Emma Chisit
I’ve spent the last few months of my spare time writing code for a model to forecast Australian elections. I’m calling the model Emma Chisit. A very basic fi...
Sentiment analysis of Q&A tweets
For a while I’ve been keen to learn how to use the Twitter API to scrape tweets. In this post I’ll do some basic sentiment analysis on tweets I scraped durin...
The Setup
In this post I answer the interview questions from The Setup, a nice website slash newsletter featuring interviews with interesting creative and technical pe...
Pitchfork and my iTunes library
I listen to a lot of music and for more than a decade now my main source of discovering new music has been Pitchfork.com, particularly since all my friends o...
Aggregating Israeli opinion polls
The Israeli electoral system is completely different to the Australian system. 120 representatives are elected to the Knesset from a single at-large electora...
Clustering Australian electorates by demographic variables
In this post we’ll use census data to cluster Australian federal electorates according to their demographic similarity. This is a good exercise for me to use...
Correcting for reliability in Australian election polls
In a post last year, I discussed the difficulties in building a model for predicting Australian elections. In the next series of posts, I want to build a few...
Tell me - what was your PhD about?
This post is an attempt for me to organise in my own head an answer to a question I’m asked all the time: what was your PhD about?
Scraping and joining Google trends searches
If you search Google trends for a term like “taylor+swift”, and set your timescale to “2004 to present”, you’ll get an interesting graph that does demonstrat...
2014
Difficulties in forecasting elections in Australia
In this post I will outline several of the difficulties involved in creating a model for forecasting Australian parliamentary elections.
Scraping Gumtree
Gumtree.com.au is a trading post website, largely used by private sellers interacting with each other off-site to sell used goods. One particularly common us...