Naive Bayesian Classifiers: How They Actually Work
Naive Bayesian classifiers are built on an assumption that’s almost always technically wrong. Here’s why it works anyway, and when it’s the right tool.
Read more →Blog
Notes on data engineering, AI and building things end to end.
Support Vector Machines Explained: The Margin That Matters
Plenty of lines can separate two classes of points. A support vector machine specifically finds the one with the widest possible margin. Here’s how and why.
Read more →
Gradient Boosting Explained: How It Actually Works
Gradient boosting gets described often and explained rarely. Here’s how it actually works, step by step, and the hyperparameters that genuinely matter.
Read more →
Vectorization vs Loops in Python for Data Processing
A for-loop over a million rows works, just slowly. Here’s vectorization vs loops in Python, why vectorized operations win, and when a loop is still the right call.
Read more →
Integral Image: What It Is and Why It’s Fast
Summing pixel values inside a rectangle, over and over, would be unbearably slow done naively. Here’s what an integral image is, and why it makes that instant instead.
Read more →
Null Hypothesis vs Alternative Hypothesis: How to Write Them
Before running a statistical test, two competing statements need to be written down clearly. Here’s null hypothesis vs alternative hypothesis, and how to write both correctly.
Read more →
Unsupervised Clustering: What It Is and How to Judge It
No labels, no correct answer to check, just raw data and a question about structure. Here’s what unsupervised clustering actually does, and how to judge if it’s good.
Read more →
L2 Regularization Formula Explained
The L2 regularization formula adds a squared-magnitude penalty to the loss function. Here’s what each term means, with a worked numeric example.
Read more →
L0 Norm and L0 Regularization Explained
L1 and L2 get all the attention, but the L0 norm is the most direct way to penalize model complexity. Here’s what it actually is, and why L1 approximates it instead.
Read more →
Table Partitioning Strategies for Large Data Tables
A query scanning a billion rows to find last week’s data is doing unnecessary work. Here’s how table partitioning strategies fix that, and how to pick the right key.
Read more →