The Mathematics Behind Machine LearningKnow the Machine series · Book 1
What if a machine could tell you a story?
Machine learning is not magic and it is not a black box. It is arithmetic: a straight line fitted through a scatter of points, a slope followed downhill, one probability multiplied by another. This book opens with a line drawn through a handful of house prices and closes with eigenvectors, maximum likelihood, and a proof that the two cost functions the whole field runs on are one principle under two assumptions.
Twenty-one chapters · 384 pages · 59 full-page infographics.
Nothing is asserted that has not been derived, and every formula is built from the one before it.
Five moves, and every chapter is one of them: take the raw facts, find what actually matters in them, build the understanding a step at a time, make the prediction, then learn from the mistake and go again. Lines and least squares. Gradients and descent. Probability, Bayes and the arithmetic of doubt. Trees, forests, eigenvectors and the SVD. Then the Python to run all of it.
Floor area →
─ least squared error
For curious minds of every age. If you can manage school algebra you can start on page one; if you already write the code, this is the derivation underneath the library call. The fifteen-year-old is why every step is spelled out. The father is why none of it is waved through.
For readers who want the mathematics explained, not waved at.
- Extent
- 384 pages, 21 chapters
- Plates
- 59 full-page infographics
- Trim
- 7 × 10 in
- Interior
- Black and white
- Binding
- Perfect bound
- Paper
- Cream / 50# white
- Paperback ISBN
- 978-1-911866-04-6
- Imprint
- Honeycomb Publications, 2026
Set in TeX Gyre Pagella. Mirrored margins for perfect binding, so the text block sits clear of the spine on every page.
Published in paperback, out now. On Amazon since September 2026, and in wholesale distribution from 18 September, which is what lets any bookshop or library order a copy.
Paperback · £14.99
Kindle and hardcover editions are not yet published.
Translation, audio and territorial rights enquiries are handled directly by the imprint.
Honeycomb Publications · East Yorkshire
honeycombscreenrights@outlook.com