homeprojectsblogcredentialsusesaboutcontact

Blog

Notes on ML, model evaluation, and what I learn building things.

Prompting Techniques: Baseline Beat Every Chain-of-X Technique

A rigorous benchmark of 9 Chain-of-X prompting techniques across 2 LLMs and 5 task types reveals that plain baseline prompting outperformed every technique on quality - while costing 2.27× less latency.

LLMPromptingBenchmarkingGroqLangChainoriginally on Namespace
Jul 20, 2026
Does Feature Engineering Still Matter in the Age of Gradient Boosting?

A controlled study comparing model performance on raw versus engineered feature sets across clinical diagnosis and financial market prediction, proving that thoughtful feature engineering consistently improves model performance.

MLFeature EngineeringXGBoostExperimentationoriginally on Namespace
Jul 14, 2026
Correlation Beats Complexity: What 20+ Models Taught Me About Learning Signal

A systematic experiment across 20+ regression and classification models with varying feature-target correlation (ρ = 0.0 to 1.0) reveals that signal quality dominates model complexity - and exposes an unexpected MAPE artefact along the way.

MLExperimentationRegressionClassificationSignaloriginally on Namespace
Jun 30, 2026
Why Accuracy Is the Most Misleading Metric in Machine Learning

Accuracy is the first metric you learn - and the one most likely to lie to your face. A structured experiment across 6 classifiers and 7 regressors exposes exactly how far accuracy can mislead you.

MLMetricsEvaluationClassificationoriginally on Namespace
Apr 11, 2026
How Linear Regression Behaves When Data Distribution Changes

A hands-on experiment: what happens to a Linear Regression model's predictions when the test distribution shifts away from training? The results are more nuanced than you'd expect.

MLExperimentationLinear Regressionoriginally on X (Twitter)
Oct 15, 2025
Why Data, Not Models, Is the Real Power Behind ML & AI

In ML, models get all the glory - but data does all the work. A controlled experiment across 5 classifiers proves that balanced, clean data outperforms complex architectures on messy data, every time.

MLDataEngineeringoriginally on Namespace
Sep 20, 2025