The statistical why behind every ML algorithm

For ML engineers and data scientists who want to understand the mathematics driving their models, not just tune the hyperparameters. Intuition first, code second.

New article every week.

Browse Articles Subscribe

Latest Articles

Likelihood: The Most Important Concept in ML That Nobody Teaches You

Foundations

Probability

Every time you call model.fit(), you’re doing maximum likelihood estimation. MSE, cross-entropy, and Poisson deviance are all negative log-likelihoods in disguise. Understanding that single fact unifies everything you know about training models.

Mar 6, 2026

Expected Value, Variance, and Why Your Loss Function Is a Statistic

foundations

probability

You minimise loss functions every day. MSE, cross-entropy, MAE, they’re the core of model training. But each one is an expected value in disguise. Understanding that connection changes how you think about every model you build.

Feb 16, 2026

Random Variables and Distributions: What Your Data Actually Is

foundations

probability

Your dataset isn’t just numbers in a CSV. It’s a realisation of random variables drawn from an unknown distribution. Until you understand that, every model you build is a guess about a process you haven’t described.

Feb 13, 2026

Why Statistics Matters Before You Touch Machine Learning

foundations

machine-learning

Everyone wants to build AI. But the real superpower isn’t knowing which algorithm to use,it’s understanding why it works.

Feb 10, 2026

Topics we’re building toward

Likelihood & MLE Bias-Variance Tradeoff Regularisation as Prior Hypothesis Testing for ML Bayesian Inference PCA & Eigendecomposition GLMs Causal Inference

Written by Godwill. MSc Health Data Science at University of Galway. Background in operations research, Bayesian modelling, and years of applied M&E work in health systems. I believe the future of AI depends on engineers who understand the statistics beneath the algorithms. More about me →

The statistical why behind every ML algorithm

Latest Articles

Likelihood: The Most Important Concept in ML That Nobody Teaches You

Expected Value, Variance, and Why Your Loss Function Is a Statistic

Random Variables and Distributions: What Your Data Actually Is

Why Statistics Matters Before You Touch Machine Learning

Topics we’re building toward

One statistical concept, explained clearly, every week