Valeriy M., PhD, MBA, CQF

Did you know the man who trained Moscow’s teachers for forty years left a bank of 1,600 arithmetic problems? Fyodor Egorov spent more than forty years at the Moscow Teachers’ Institute and ran it from 1895. His son Dmitri became one of the founders of the Moscow school of mathematics.

A pupil has dropped a zero or misread a digit; reconstruct the right result from his faulty working. Railway shares, the silver in a ruble, noon in Riga, Petersburg and Saratov. Poods, versts and sazhens beside meters and kilograms. 149 pages. Pairs with Kiselev’s Arithmetic.

Did you know the man who trained Moscow’s teachers for forty years left a bank of 1,600 arithmetic problems? Fyodor Egorov spent more than forty years at the Moscow Teachers’ Institute and ran it from 1895. His son Dmitri became one of the founders of the Moscow school of mathematics.

This is the problem book the father built for the daily work of the classroom. 1,600 graded problems in whole-number arithmetic. Answers to 1,245, every one checked. What makes it different: Four whole sections on how answers change. What happens to a product when one factor grows?

In 1772 an eye operation left Leonhard Euler almost completely blind. He could still do the mathematics. He could not read it or write it. So he needed someone. Not a copyist — someone who could follow him. Euler turned to his old friend Daniel Bernoulli in Basel.

Then the application of algebra to geometry — trigonometry, conic sections, and the foundations of the calculus. That trilogy is where Russian school mathematics begins. Kiselev comes after it. The Soviet school comes after Kiselev. Part I — Algebra, 1820.

In 1772 an eye operation left Leonhard Euler almost completely blind. He could still do the mathematics. He could not read it or write it. So he needed someone. Not a copyist — someone who could follow him. Euler turned to his old friend Daniel Bernoulli in Basel.

Then Euler died in 1783, and the assistant became the scientist. Academician that same year. Professor of mathematics. And from 1800 until his own death in 1826, permanent secretary of the Imperial Academy of Sciences. There is a theorem named after him. A sequence of numbers too.

In 1772 an eye operation left Leonhard Euler almost completely blind. He could still do the mathematics. He could not read it or write it. So he needed someone. Not a copyist — someone who could follow him. Euler turned to his old friend Daniel Bernoulli in Basel.

A lot of data scientists in 2026 still either skip EDA or do it wrong. They think EDA is about making pretty charts. It is not. EDA is not a gallery of polished plots. EDA is not a box-ticking exercise before modeling. EDA is not “let’s run a few charts and move on.”

So before you train anything, investigate the world. Not casually. Not cosmetically. Forensically. Pretty pictures are optional. What is not optional is understanding what the data really is, how it came to exist, and how it can mislead you. That is real EDA.

A lot of data scientists in 2026 still either skip EDA or do it wrong. They think EDA is about making pretty charts. It is not. EDA is not a gallery of polished plots. EDA is not a box-ticking exercise before modeling. EDA is not “let’s run a few charts and move on.”

* Which features would not exist at prediction time? * Where are the regime shifts, bad joins, label issues, and hidden biases? That is what EDA is for. EDA is your first forensic examination of the data-generating process. It helps you uncover things that destroy models in the real world:

A lot of data scientists in 2026 still either skip EDA or do it wrong. They think EDA is about making pretty charts. It is not. EDA is not a gallery of polished plots. EDA is not a box-ticking exercise before modeling. EDA is not “let’s run a few charts and move on.”

@predict_addict · Jan 16 Reading Andrei Markov’s Calculus of Probability feels like drinking from the holy source of probability. After the early spark from Pascal and Fermat, probability theory in Western Europe largely stalled—fragmented, ad-hoc, and philosophically confused for centuries.

This should have been the standard probability textbook everywhere — but it wasn’t, largely because it was never properly translated or absorbed outside Russia.

@predict_addict · Jan 16 Reading Andrei Markov’s Calculus of Probability feels like drinking from the holy source of probability. After the early spark from Pascal and Fermat, probability theory in Western Europe largely stalled—fragmented, ad-hoc, and philosophically confused for centuries.

Markov is one of the clearest embodiments of that restart. What strikes you immediately is not just the mathematics, but the clarity of thought and writing: no mysticism, no hand-waving, no pretending probability is “obviously” a number in all situations, no cargo-cult formalism.

@predict_addict · Jan 16 Reading Andrei Markov’s Calculus of Probability feels like drinking from the holy source of probability. After the early spark from Pascal and Fermat, probability theory in Western Europe largely stalled—fragmented, ad-hoc, and philosophically confused for centuries.

Forget the faux pas “calculus” marketing product. US universities invented it to charge massive fees and push shallow, overpriced books like Spivak. The real subject is mathematical analysis — the way it was taught for decades in the USSR and China. The iconic book:

Fikhtengol’ts — Foundations of Mathematical Analysis, Volume I 🔥🏆🚀 Lectures by the professor whose students included Sergei Sobolev, Leonid Kantorovich (Nobel Prize in Economics), Pelageya Polubarinova-Kochina and Viktor Ambartsumian.

The most rigorously written introduction to school algebra has six chapters and 126 numbered paragraphs. Signed numbers. Polynomials. Algebraic fractions. Equations of the first degree in one, two, and three unknowns. Square roots. Quadratic equations.

Forget the faux pas “calculus” marketing product. US universities invented it to charge massive fees and push shallow, overpriced books like Spivak. The real subject is mathematical analysis — the way it was taught for decades in the USSR and China. The iconic book:

Fikhtengol’ts — Foundations of Mathematical Analysis, Volume I 🔥🏆🚀 Lectures by the professor whose students included Sergei Sobolev, Leonid Kantorovich (Nobel Prize in Economics), Pelageya Polubarinova-Kochina and Viktor Ambartsumian.

How Muscovite arithmeticians found “the money in a heap”: Count in 3s — for each coin left over, write 70. Count in 5s — write 21 per coin. Count in 7s — write 15 per coin. Add up. Subtract 105 until you can’t. Remainders 2, 3, 2 → 233 → 23 coins.

Applied Conformal Prediction is now on Amazon in print, in full colour. 📘 Colour paperback: www.amazon.com/dp/18... 📕 Luxury colour hardcover: www.amazon.com/dp/18... Your model says 0.87. How often is it right when it says 0.87? Conformal prediction answers that.

→ Distribution shift, time series and online adaptive methods → Conformal risk control and predictive distributions for real decisions → Vision and language models, deployment monitoring, and four end-to-end case studies

Applied Conformal Prediction is now on Amazon in print, in full colour. 📘 Colour paperback: www.amazon.com/dp/18... 📕 Luxury colour hardcover: www.amazon.com/dp/18... Your model says 0.87. How often is it right when it says 0.87? Conformal prediction answers that.

It wraps any model, from linear regression to gradient boosting to deep networks, and returns prediction sets and intervals with finite-sample coverage guarantees. The assumptions behind them are stated, and you can test them.

The day has come. Thank you to all who supported this book, especially the early supporters. Applied Conformal Prediction is now on Amazon in colour paperback and luxury colour hardcover. The future of uncertainty quantification is bright, the future is #conformalprediction

He was refused his degree for skipping theology lectures. Then he went to Paris and found the divergence theorem. Mikhail Ostrogradsky studied at Kharkov. A reactionary professor accused him of not attending lectures on Christian doctrine.

Yushkevich argues that Green’s 1828 work does not contain it, and that Poisson presented a similar formula in 1828, after having read Ostrogradsky’s manuscript. Priority in mathematics is rarely simple, and special cases go back further. But the story is extraordinary. It is in A. P.

He was refused his degree for skipping theology lectures. Then he went to Paris and found the divergence theorem. Mikhail Ostrogradsky studied at Kharkov. A reactionary professor accused him of not attending lectures on Christian doctrine.

On the minister’s orders his teacher was dismissed, and Ostrogradsky was denied the degree he had earned. He left for Paris in 1822.