A graduate reading list with zero tuition

What it does Carl McBride Ellis’s repository is a curated index of 194 machine-learning books and papers that cost nothing. The README sorts them into 18 categories—from EDA and optimization to causal inference and LLMs—and marks each entry as either a PDF or an HTML book. It is essentially glue code: a well-organized reading list built from publisher-sanctioned free PDFs and arXiv links rather than original software.

The interesting bit The list mixes canonical graduate texts—Bishop’s Pattern Recognition and Machine Learning, Hastie & Tibshirani’s The Elements of Statistical Learning, Boyd & Vandenberghe’s Convex Optimization—with newer niches like conformal prediction and anomaly detection. Because it relies on author- or publisher-hosted files and arXiv, the links are (in theory) less legally precarious than the usual shadow-library PDFs.

Key highlights

• 194 resources spanning 18 categories, from mathematics for ML to time-series forecasting
• Heavy hitters included: ISLP, ESL, Bayesian Data Analysis, Murphy’s Probabilistic Machine Learning series, and Gelman & Hill’s Regression and Other Stories
• Simple visual key: :orange_book: for PDFs, :earth_africa: for HTML books
• Covers underserved topics often skipped in standard coursework: recommender systems, conformal prediction, causal inference
• Explicitly limited to legally free material—no pirated uploads

Caveats

• Pure static markdown; there is no search, tagging beyond the 18 headings, or link-rot monitoring
• Some categories are barely populated—anomaly detection lists a single book, feature engineering and recommender systems only two each
• Small rough edges like the R realted typo hint at a solo, low-friction maintenance style

Verdict A valuable bookmark for self-taught practitioners and students who want canonical references without paying canonical prices. Not useful if you need executable code, tutorials, or an interactive curriculum.