Applied Conformal Prediction is a comprehensive and practical guide to one of the most powerful and rapidly evolving frameworks in machine learning: Conformal Prediction (CP).

Written by Valery Manokhin, who completed his PhD under Vladimir Vovk, the creator of Conformal Prediction, and has been one of its most prominent advocates for years, this book reflects deep expertise and commitment to the field. Manokhin's widely followed "Awesome Conformal Prediction" repository and his contributions to the global CP community have helped fuel its meteoric rise in research and industry.

Conformal Prediction is quickly becoming a must-have skill for anyone working in high-stakes, production-level AI systems. It provides rigorous, model-agnostic methods for quantifying uncertainty and constructing statistically valid prediction sets with guaranteed coverage. Unlike many traditional approaches, CP offers finite-sample guarantees without requiring unrealistic assumptions.

This book begins with the philosophical and mathematical origins of CP and walks you through its key components: exchangeability, nonconformity scores, prediction regions, inductive and adaptive variants, and beyond. It then explores cutting-edge research on:

• Classification
• Classifier calibration
• Regression
• Time Series and Forecasting (e.g., EnbPI, blockwise CP)
• Deep Learning Integration (NLP, CV, transformers)
• Weighted CP for covariate shift
• Software tools
• And much more

Whether you're a practitioner building risk-sensitive systems or a researcher exploring the limits of statistical inference, Applied Conformal Prediction is your definitive resource.

Preorder now to lock in the lowest price. You'll get early access to chapters as they’re released, and all future updates will be included. The price will increase significantly as more chapters are added.

Mastering CatBoost: The Hidden Gem of Tabular AI (early access)

By Valeriy Manokhin, PhD, MBA, CQF

“CatBoost is not just underrated—it’s objectively better.”

This book shows you why, with the science and the code to prove it.

Preorder now and lock in lifetime access.

As the content continue to grow, if you find value in it and want to support the project - you are welcome to contribute whatever it is woth to you ❤️.

Why CatBoost

There’s a preponderance of scientific evidence that CatBoost consistently and significantly (20%+ according to TabArena) outperforms XGBoost, LightGBM on real-world tabular data.

It's faster in inference, easier to tune, and built from the ground up for categorical features—without the usual preprocessing hacks.

Despite this, CatBoost remains one of the most underused tools in machine learning. This book fixes that.

Backed by research, benchmarks, and production experience

Practical, readable, hands-on for working data scientists

Linked to the open-source repo: Awesome CatBoost

What You’ll Learn

• Core architecture: how CatBoost works under the hood
• Hands-on modeling: end-to-end tabular ML pipelines
• Categorical encoding: no more label/one-hot hacks
• Overfitting detection: built-in, automated safeguards
• Evaluation strategies: cross-validation the CatBoost way
• Interpretability: SHAP, feature importance, monotonic constraints
• Bonus: Time series with CatBoost + quantile & uncertainty modeling using Conformal Prediction

Scope & Depth: More than Just Boosters

• Mastering CatBoost covers:
• Not just classification, but regression, ranking, time series, and even quantile/uncertainty models
• Deep dive into categorical feature handling (one of CatBoost’s many advantages)
• Native overfitting detection, monotonic constraints, and interpretability tools all built-in and tuned for tabular workflows

Under-the-Hood Architecture & Scientific Advantages

• Mastering CatBoost delves into:
• Ordered boosting, symmetric trees, and smoothed target statistics — explaining why CatBoost handles categorical variables without leakage
• Scientific benchmarks consistently show CatBoost outperforming XGBoost and LightGBM on real-world tabular datasets
• Includes newer capabilities like GPU optimizations, quantization, and ONNX export

Interpretability & Safeguards

• Native overfitting detection, eliminating guesswork
• Built-in per-feature importance, interaction, and partial dependence tools
• Monotonic constraints tuned specifically for CatBoost internals

The Verdict

• Mastering CatBoost goes far beyond:
• In technical depth (architecture + categorical handling)
• Applied scope (classification, regression, ranking, forecasting)
• Deployment readiness (quantization, ONNX, real-world pipelines)
• Support materials (Awesome_CatBoost repo, notebooks, domain-specific chapters)

Who Is This For?

This book is designed for:

• Machine learning engineers using tabular datasets
• Data scientists tired of endless hyperparameter tuning
• Students or researchers who’ve hit limits with XGBoost or sklearn
• Practitioners who want to move fast from data to insight

If you like fast iteration, fewer bugs, and state-of-the-art tabular models, this book is for you.

What You Get

Instant access to the book — start reading published chapters immediately.

Free updates — including new chapters, bug fixes, and bonus content.

Exclusive access to the private Discord community — connect with fellow readers, get additional materials, early bonuses, special discounts, and join live events with the author.

✍️ About the Author

Written by Valeriy Manokhin, PhD, MBA, CQF — a seasoned AI, conformal prediction and forecasting expert, data scientist, and machine learning researcher with publications in top academic journals.

Valeriy has advised both startups and large enterprises, helping them build and rebuild forecasting systems at scale. He has led successful forecasting initiatives for global organizations — including winning competitive tenders from multinational companies, outperforming major consulting firms like BCG and specialized AI startups focused on forecasting. He has delivered production-grade solutions for industry leading Fortune 500 companies/

His methods have driven multimillion-dollar business impact, and his training programs have reached professionals in over 40 countries. His book Mastering Modern Time Series Forecasting is now used in more than 100+ countries and has become a #1-ranked title in Machine Learning, Forecasting, and Time Series across major platforms.

Trusted By and Taught To

Valeriy’s expertise is trusted by leaders at:

Amazon, Apple, Google, Meta, Nike, BlackRock, Morgan Stanley, Target, NTT Data, Mars Inc., Lidl, Publicis Sapient, and more.

His frameworks are followed by professionals from:

University of Chicago, KTH (Sweden), UBC (Canada), DTU (Denmark), and other world-class institutions.

Students include:

VPs of Engineering, AI Leads, Principal & Lead Data Scientists, ML Engineers, Consultants, Professors, Founders, Researchers, and PhD students.

Also by the Author

Mastering Modern Time Series Forecasting 

The book trusted by data science leaders in 100+ countries. Unlock the toolkit behind today’s most powerful forecasting systems.

Learn more → MasteringModernTimeSeriesForecasting

⚡ Ready to Master the Best Tabular Model in ML?

CatBoost isn’t just another gradient booster.

It’s the most underappreciated breakthrough in machine learning—and you’re about to master it.

Grab your copy now and start building faster, better models with less tuning.