Reading Guide
The concept pages teach the main path. Full-book reading is not required. Read to repair a specific gap, then close the source and solve a changed case. The existing local collection supplies the two backbone books; external primary sources add alternate exercises and a precise interval formula.
| Stage | Selective reading | Reading task | Evidence afterward |
|---|---|---|---|
| Models and counting | Introduction to Probability: sample spaces, counting, complements | Compare uniform ordered outcomes with a nonuniform summary | Explain practice A1 without a formula lookup |
| Conditioning | Same book: conditional probability, Bayes, independence | Draw a tree and turn it into a count table | Solve A2 and explain why duplicate evidence is not a second update |
| Random variables | Same book: PMF/CDF, named discrete distributions | Identify fixed trial counts, stopping, and replacement | Explain why B1 and B2 need different models |
| Expectation and dependence | Same book: expectation, LOTUS, indicators, covariance | Derive one expectation without the full distribution | Derive retry counts and empty-server expectation |
| Proof reinforcement | Mathematics for Computer Science: random variables, expectation, deviation bounds | Identify every assumption in a variance or Chebyshev argument | Explain copied-row variance and where independence was used |
| Averages | Pishro-Nik's CLT section | Separate exact variance algebra from limiting distribution | Interpret Lab C without claiming a universal sample threshold |
| Intervals | NIST: confidence intervals for a proportion | Locate Wilson, Wald, and exact-binomial distinctions | Explain the zero-count result and enumerate coverage |
| Independent extension | Harvard Stat110 strategic practice and homework | Choose two topic exercises, then one mixed exercise without its solution | Submit model, derivation, limitation, and correction if needed |
The generated local book chunks sometimes combine several textbook sections. These links open book indexes intentionally; use the topic names above and each book's contents. A label such as "LOTUS" must not pretend that an unrelated generated chapter is the exact section. Different editions can change page and exercise numbering, so this guide assigns topics and outputs rather than invented page references.
For Harvard exercises, use the source's own solutions only after an attempt. The module's assessment uses newly authored problems and its own rubric; completing a source exercise does not automatically validate this course's assessment standard.
Continuous models and inference extension
The required inference extension guide adds targeted primary sources for concepts 16–20. Its local practice and topic readings are included in the revised workload; extra full courses remain optional.