Library

The goal of library is to curate the most useful and interesting resources I’ve relied on over time.

Each item (hopefully) expands to include personal notes, summaries, and commentary. The external link (↗) opens the original source.

Books

All of Statistics — Wasserman ↗
  • A concise but mathematically rigorous introduction to probability and statistical inference.
  • Covers measure-theoretic probability (lightly), estimation, hypothesis testing, regression, nonparametrics, and bootstrap.
  • Appropriate for readers with multivariable calculus and some linear algebra.
  • Strong emphasis on theory with compact proofs; useful as a bridge between undergraduate probability and graduate-level statistics.
  • Best for readers who want mathematical clarity without the length of a traditional 800+ page text.
Statistical Inference — Casella Berger ↗
  • A canonical graduate-level treatment of mathematical statistics.
  • Deep coverage of likelihood theory, sufficiency, exponential families, estimation theory, hypothesis testing, and asymptotics.
  • Emphasizes formal derivations and theoretical rigor.
  • Suitable for mathematically mature readers (real analysis recommended).
  • Often used in first-year PhD statistics programs.
Elements of Statistical Learning ↗
  • A comprehensive, theory-driven treatment of statistical learning methods.
  • Covers linear models, splines, kernel methods, SVMs, boosting, trees, ensemble methods, model assessment, and high-dimensional statistics.
  • Assumes comfort with linear algebra, multivariate calculus, and some probability.
  • Emphasizes geometry and statistical principles over code.
  • Best suited for readers who want to understand the theoretical foundations of modern machine learning.
An Introduction to Statistical Learning ↗
  • A more accessible and applied counterpart to Elements.
  • Focuses on conceptual understanding and implementation.
  • Includes worked examples, labs, and exercises.
  • Available in R (ISLR) and Python (ISLP) editions.
  • Ideal for practitioners transitioning from classical statistics to applied machine learning.
Forecasting: Principles and Practice — Hyndman ↗
  • A practical and modern introduction to time series forecasting in R.
  • Covers classical methods (ETS, ARIMA), regression with ARIMA errors, hierarchical forecasting, and forecast evaluation.
  • The 3rd edition starts using tsibble and fable instead of the forecast package.
  • There is a python version: FPP: The Pythonic Way
Trustworthy Online Controlled Experiments — Kohavi ↗
  • The definitive book on A/B testing.
  • Covers experimentation design, metrics, guardrails, statistical pitfalls, and organizational challenges.
  • Draws heavily from Microsoft and industry case studies.
  • Emphasizes experimentation infrastructure and culture, not just statistical testing.
  • Essential for anyone working in product analytics or growth experimentation.
Machine Learning with PyTorch and Scikit-Learn — Raschka ↗
  • A hands-on, implementation-oriented guide to modern ML in Python.
  • Covers classical ML (via scikit-learn) and deep learning (via PyTorch).
  • Includes end-to-end workflows: preprocessing, model training, tuning, and deployment considerations.
  • Strong focus on reproducible experimentation and practical coding patterns.
  • Appropriate for practitioners who want to move beyond theory into production-capable workflows
Bayesian Data Analysis — Gelman ↗
  • A comprehensive and rigorous treatment of Bayesian modeling.
  • Covers probability foundations, hierarchical models, MCMC, model checking, and applied case studies.
  • Strong emphasis on multilevel modeling and posterior predictive checks.
  • Requires mathematical maturity; used in graduate-level coursework.
  • A definitive reference for applied Bayesian statistics.
Data Analysis Using Regression and Multilevel/Hierarchical Models — Gelman ↗
  • Focuses on regression modeling with partial pooling and hierarchical structures.
  • Bridges classical and Bayesian perspectives.
  • Emphasizes modeling strategy, diagnostics, and interpretation.
  • More applied and less mathematically dense than Bayesian Data Analysis.
Doing Bayesian Data Analysis — Kruschke ↗
  • The puppy book!
  • A highly accessible introduction to Bayesian inference.
  • PyMC-based adaptations are available.
  • Ideal for readers new to Bayesian thinking.
Ace the Data Science Interview — Singh ↗
  • A broad survey of topics frequently tested in DS interviews.
  • Covers SQL, product metrics, experimentation, probability, machine learning, and case questions.
  • More strategic and tactical than theoretical.

Videos

Neural Networks: Zero to Hero — Andrej Karpathy ↗
  • Great practical build-up from micrograd to GPT-style training.
  • Created by the co-founder of OpenAI.

Web

StrataScratch ↗
  • LeetCode for data science.
  • Good way to stay sharp on data structures, pandas, pyspark, SQL, and more.
LeetCode ↗
  • The ubiquitous platform for data structure practice.

Papers

For a reminder that things are more modern than you might expect.

On Computable Numbers, with an Application to the Entscheidungsproblem (1936) — A. M. Turing
  • The turing machine itself.
Foundations of the Theory of Probability (1933) — Andrey Kolmogorov
  • Axioms of probability.
Regression Shrinkage and Selection via the Lasso (1996) — Robert Tibshirani
  • The LASSO!
Attention Is All You Need (2017) — Ashish Vaswani et al.
  • Transformer models!
On the Mathematical Foundations of Theoretical Statistics (1922) — R. A. Fisher
  • Maximum Likelihood Estimation!