The maths behind the methods, and the edges they break at.
I write here about the mathematics I find worth setting down.
Local time
Monday 13 Jul
London · GMT+1
Using principal component analysis and manifold-learning methods to extract latent structure from high-dimensional electricity demand data and discovering how much information the curse of dimensionality quietly takes from you.
End-to-end data science workflow predicting SpaceX Falcon 9 landing success. Covers API data collection, web scraping, SQL exploration, geospatial visualization with Folium, and supervised classification (Logistic Regression, SVM, Decision Tree, KNN) achieving 83.33% accuracy.
Comprehensive Power BI dashboard built from three raw CSV datasets. Full BI workflow — ETL with Power Query, star schema data model, and interactive dashboard analysing sales performance, profitability, and customer loyalty across global markets.
Hackathon-winning platform that bridges disciplinary silos at universities. Students post collaboration requests via voice or text, processed by ElevenLabs and Google Gemini for AI-powered tagging and collaborator matching. Won MLH Best Use of ElevenLabs.
Estimate π by throwing darts, ride the Galton board down to the bell curve, out-think the Monty Hall problem, and read correlation by eye. Five interactive machines — no libraries, just maths.
Enter the playgroundWhy high-dimensional data breaks the statistics you were taught, and what you can actually do about it.
How three simple integrators behave on real problems: exponential growth, a stiff Van der Pol oscillator, and three-body figure-eight orbits.
How a single question in a 1901 letter collapsed Frege's project to build mathematics out of pure logic.