Read
Short, practical reads on AI and data — from first principles to RAG, agents, MCP and AI security. Free, no sign-in.
55 guides in Machine learning
Using plots of performance against data size and model settings to diagnose models and decide what to try next.
Machine learning 1 min read 16 Aug 2026 For people
How ROC curves show classifier performance across thresholds, what AUC measures, and when precision-recall curves are better.
Machine learning 1 min read 15 Aug 2026 For people
MAE, RMSE, MAPE and R² explained: what each measures and how to choose the right one for your problem.
Machine learning 1 min read 14 Aug 2026 For people
The difference between choosing one class from many and assigning several labels at once, with methods and metrics for each.
Machine learning 1 min read 13 Aug 2026 For people
Practical ways to create informative features from raw data: dates, text, aggregates, interactions and domain knowledge.
Machine learning 1 min read 12 Aug 2026 For people
Turning text into features with bag-of-words, TF-IDF and n-grams, and when embeddings are a better choice.
Machine learning 1 min read 11 Aug 2026 For people
Practical techniques that make neural networks train well: initialisation, normalisation, dropout, optimisers and schedules.
Machine learning 1 min read 10 Aug 2026 For people
What activation functions do in neural networks, and the strengths of ReLU, sigmoid, tanh, GELU and softmax.
Machine learning 1 min read 9 Aug 2026 For people
How loss functions define what a model optimises, and common choices for regression, classification and ranking.
Machine learning 1 min read 8 Aug 2026 For people
Neural networks that learn to compress and reconstruct data, and their uses in dimensionality reduction, denoising and anomaly detection.
Machine learning 1 min read 7 Aug 2026 For people
How agents learn from rewards through trial and error: states, actions, policies, exploration and real-world uses.
Machine learning 1 min read 6 Aug 2026 For people
A lightweight alternative to A/B testing that learns which option works best while still serving users.
Machine learning 1 min read 5 Aug 2026 For people
Why predictive models don't answer 'what if' questions, and the methods used to estimate causal effects from data.
Machine learning 1 min read 4 Aug 2026 For people
Why models should say how sure they are, and methods for quantifying uncertainty in predictions.
Machine learning 1 min read 3 Aug 2026 For people
Why gradient-boosted trees often beat neural networks on spreadsheet-style data, and when deep learning is worth trying.
Machine learning 1 min read 2 Aug 2026 For people