Profile photo

Hej there

Tech, side projects, and life in the PNW. Written by a data scientist who splits time between pistes and debugging.

Decision Trees and Ensemble Methods

From single decision trees to random forests and gradient boosting - how combining models reduces error and when to use each approach.

Hypothesis Testing, Confidence Intervals and Power

The statistical decision-making toolkit - hypothesis tests, p-values, confidence intervals, power, and sample size planning.

Linear Regression

The simplest predictive model - how it works, how to fit it, what can go wrong, and how to fix it.

Model Evaluation

Loss functions, metrics, cross-validation, and diagnostic curves - how to measure whether your model actually works.

Regularization and Feature Selection

Preventing overfitting and choosing the right features - bias-variance trade-off, Ridge, Lasso, and feature selection methods.

Probability Basics

The rules of uncertainty: axioms, conditional probability, independence, and Bayes’ theorem.

Probability Distributions

The shapes data takes - random variables, PMF, PDF, CDF, and the key distributions you need to know.

The Central Limit Theorem

Why the normal distribution dominates statistics - sampling distributions, the law of large numbers, and the CLT.

Population, Samples and Descriptive Statistics

The starting point for statistics: what we measure, how we summarise it, and why the sample formula divides by n-1.

Building This Blog: Hugo + PaperMod + Cloudflare Pages

I wanted a blog that’s fast, cheap to host, and easy to maintain. After evaluating a few options, I landed on Hugo with the PaperMod theme, deployed to Cloudflare Pages. Here’s the full breakdown. The stack Static site generator: Hugo — fast builds, great templating, large ecosystem Theme: PaperMod — clean, minimalist, fast, well-maintained Hosting: Cloudflare Pages — free tier, global CDN, automatic deploys from GitHub Comments: Giscus — powered by GitHub Discussions Domain: Cloudflare Registrar (at-cost pricing, ~$10/year) Why Hugo? I considered Next.js, Astro, and Jekyll before settling on Hugo: ...