Syllabus
STAT 432 | Basics of Statistical Learning
The assessment weights and weekly topic sequence below are current. Items marked TBA will be posted before they take effect.
Course Information
- Instructor: Ruoqing Zhu
- Teaching assistant: Chenyu Zhang
- Instructor email: rqzhu@illinois.edu
- Teaching assistant email: chenyuz9@illinois.edu
- Meeting days: Tuesday and Thursday
- Meeting time: 12:30 to 1:50 p.m.
- Classroom: 2039 Campus Instructional Facility
- Instructor office hours: Friday, 3:00 to 4:00 p.m.
- Instructor office: 137 CAB
- Teaching assistant office hours: Monday, 2:00 to 4:00 p.m., on Zoom
Thursday classes focus on mathematical foundations, selected derivations, examples, and practical issues. Tuesday normally begins with a 50- to 60-minute discussion session about the previous Thursday’s topic. The session moves from Group Discussion to Present and Challenge. A short lecture may follow if time permits. On quiz Tuesdays, the 30-minute quiz takes place before the discussion.
Course Description and Learning Goals
STAT 432 introduces the basic ideas and methods of statistical learning. The course connects mathematical reasoning with practical work in R or Python. Students will also learn through weekly questions, group discussion, and explanation of statistical ideas to classmates.
By the end of the course, students should be able to:
- explain the main ideas behind widely used regression, classification, and unsupervised-learning methods;
- connect mathematical results to model behavior in practice;
- fit, tune, evaluate, and interpret statistical-learning models in R or Python;
- reason about model complexity, prediction error, and uncertainty;
- use simulation, resampling, and numerical optimization to study statistical methods;
- communicate a statistical argument in writing, code, and discussion;
- organize reproducible work with GitHub; and
- use AI coding agents responsibly and check the work they produce.
Prerequisites
Students should have completed a course that covers linear regression, such as STAT 420 or STAT 425. Working knowledge of probability and statistics at the STAT 400 level and basic linear algebra at the MATH 415 level, or equivalent preparation, are assumed. Students should also have basic experience writing code in either R or Python.
Topics and Weekly Schedule
The table lists the Tuesday and Thursday meetings in each calendar week. Tuesday normally discusses the topic introduced on the previous Thursday. The homework and discussion question for that topic are both due by 11:59 p.m. Central Time on the Sunday before that Tuesday. On quiz Tuesdays, the 30-minute quiz takes place before the discussion.
Detailed lecture materials are currently posted through Week 5, and homework materials are available for Weeks 1 through 5. Additional materials will be added after review.
| Week number | Tuesday | Thursday |
|---|---|---|
| 1 | Aug 25: Course introduction | Aug 27: Course setup, GitHub, AI tools, skills, and discussion-question procedures |
| 2 | Sep 1: Guided first discussion | Sep 3: Training and Test Error |
| 3 | Sep 8: Discussion of Week 2 | Sep 10: Ridge Regression and Optimization |
| 4 | Sep 15: Quiz 1, covering Weeks 1 to 2, then discussion of Week 3 | Sep 17: Lasso and Variable Selection |
| 5 | Sep 22: Discussion of Week 4 | Sep 24: K-Nearest Neighbors |
| 6 | Sep 29: Discussion of Week 5 | Oct 1: Classification Error and Evaluation |
| 7 | Oct 6: Quiz 2, covering Weeks 3 to 5, then discussion of Week 6 | Oct 8: Kernel Smoothing and Local Regressions |
| 8 | Oct 13: Discussion of Week 7 | Oct 15: Support Vector Machines |
| 9 | Oct 20: Discussion of Week 8 | Oct 22: Trees and Random Forests |
| 10 | Oct 27: Quiz 3, covering Weeks 6 to 8, then discussion of Week 9 | Oct 29: Boosting |
| 11 | Nov 3: Discussion of Week 10 | Nov 5: Unsupervised Learning Algorithms |
| 12 | Nov 10: Discussion of Week 11 | Nov 12: Observational Data and Causal Inference |
| 13 | Nov 17: Quiz 4, covering Weeks 9 to 11, then the final discussion of Week 12 | Nov 19: Selected Topics in AI |
| 14 | Nov 24: Fall Break, no class | Nov 26: Fall Break, no class |
| 15 | Dec 1: Final project presentations | Dec 3: Final project presentations |
| 16 | Dec 8: Final project presentations | Dec 10: Reading Day, no class |
Dates follow the Fall 2026 Academic Calendar.
The schedule may be adjusted as the course develops. Changes will be announced in class and posted on the course website.
Books and Course References
The lecture notes are the main guide for the course. The following books are supplemental references, and students are not expected to purchase all of them.
- SMLR: Ruoqing Zhu, Statistical Learning and Machine Learning with R. These online notes provide additional mathematical explanations and examples.
- ISL: Gareth James, Daniela Witten, Trevor Hastie, and Robert Tibshirani, An Introduction to Statistical Learning. R and Python editions are available from the book website.
- ESL: Trevor Hastie, Robert Tibshirani, and Jerome Friedman, The Elements of Statistical Learning. This is a more advanced reference.
How the Course Works
Beginning with Week 2, the course normally follows this cycle:
- Thursday lecture: study the mathematical foundations, selected derivations, examples, and practical issues.
- After the lecture: finish the remaining notes, complete the weekly homework, and submit one discussion question through GitHub.
- Monday: the instructor selects a set of 12 questions submitted by students to form the question pool for Tuesday.
- Tuesday class: four to eight questions are posted. Students first work through them in groups using any available tools. The class then enters Present and Challenge. Assigned teams present without notes, computers, AI tools, or outside assistance. Students from other teams may challenge an answer, and the class rates presenting teams and challengers. A short lecture may follow if time permits. On scheduled quiz Tuesdays, a 30-minute quiz takes place before the discussion.
The Week 1 lecture Discussion Questions: Procedures and Rules explains the submission and discussion process in detail.
Assessment and Grading
| Component | Weight |
|---|---|
| Weekly homework | 15% |
| Weekly GitHub discussion questions | 10% |
| Classroom discussion participation | 25% |
| Four quizzes | 25% |
| Final project | 25% |
| Total | 100% |
Weekly Homework
Weekly homework is required for Weeks 1 through 12 unless it is explicitly waived. Homework numbering begins immediately with Homework 01 in Week 1. Each homework is evaluated as Complete or Incomplete rather than being scored for correctness. Its purpose is to help students check their understanding of the mathematics, computation, interpretation, and practical choices introduced that week.
Many questions are adapted from previous STAT 432 homework, and solutions are available. Students should first identify the main statistical idea and think about how to begin. They may then work independently, use an AI coding agent, consult the solution, or combine these approaches. Students remain responsible for checking the work and being able to explain it.
Submit the final report to the corresponding homework/week-NN/ folder in your stat432-fall2026 GitHub repository. Name the file homework-NN.pdf or homework-NN.html, using the same two-digit week number. The report must be a PDF or a self-contained HTML file and must not exceed 10 pages. For an HTML report, the limit is 10 pages when printed from a browser using standard settings.
Each weekly page provides a download named stat432-homework-week-NN.zip. In your local stat432-fall2026 repository, open the existing homework/ folder and extract the ZIP there. The archive creates week-NN/, so the materials appear in homework/week-NN/. This folder includes homework-NN.qmd, solutions-NN.qmd, and any required data or figures. Render homework-NN.qmd to produce homework-NN.html or homework-NN.pdf. Only that final numbered HTML or PDF report is required for submission.
The homework for a Thursday topic is due by 11:59 p.m. Central Time on the following Sunday, before that topic is discussed on Tuesday. The corresponding weekly page gives the exact date. The file version visible on GitHub is the submitted version, and its GitHub timestamp determines whether it is on time.
Each student may use a two-day extension without penalty for up to three homework assignments during the semester. For an extended homework, the Sunday deadline becomes Tuesday at 11:59 p.m. Central Time. A submission shown on GitHub after the regular Sunday deadline but no later than the extended Tuesday deadline uses one of the three extensions. These extensions apply only to homework, not to discussion questions.
A homework is Complete when the required file is visible in the correct GitHub location by the applicable deadline, opens correctly, follows the format and page-limit requirements, and contains a substantive attempt at every required question. Correctness is not graded. An Incomplete homework earns 0 points for that assignment.
A homework submitted after its applicable deadline is Incomplete and receives 0 points. After all three extensions have been used, a homework submitted after the regular Sunday deadline is late. Any homework submitted after the extended Tuesday deadline is late. The lowest weekly homework score is dropped when the Weekly Homework component is calculated.
The six Week 1 foundation questions make up Homework 01. They follow the same completion, file-format, deadline, extension, and late-submission rules as later homework.
Weekly Discussion Questions
Students submit one discussion question for each instructional week unless the question is explicitly waived. These submissions make up 10% of the course grade.
- Use a GitHub repository named exactly
stat432-fall2026. - Save the question as
discussion/week-NN-question.md, using a two-digit week number. - Begin with the required five-line YAML frontmatter. Use your Illinois NetID as the
author; a real or full name and GitHub username are optional. - Keep the prose body to at most 200 words. Fenced code and LaTeX mathematics do not count toward this limit.
- Submit by Sunday at 11:59 p.m. Central Time after the Thursday lecture.
- Proofread the complete file, commit it, push it, and verify the version shown on GitHub.
The question must reflect the student’s own thinking. An AI coding agent may improve the formatting or clarity, but it may not invent the student’s question, confusion, reasoning, result, or opinion. The file version visible on GitHub is the submitted version. If its GitHub timestamp is after the Sunday deadline, the submission receives 0 points. No late-submission exceptions are allowed.
See Discussion Questions: Procedures and Rules for the complete format and submission instructions.
Tuesday Discussion Participation
The instructor selects 12 submitted questions to form the weekly pool, and four to eight are posted during the Tuesday session. The session discusses the topic taught on the previous Thursday and usually lasts 50 to 60 minutes. Students first enter Group Discussion to work through the posted questions and prepare explanations. The class then enters Present and Challenge. Presenting teams are assigned randomly to questions. A challenger is a student from another team who raises an important question, identifies a limitation, or proposes a different argument. A short lecture may follow if time permits.
The classroom discussion component makes up 25% of the course grade and is based entirely on participation during the Tuesday sessions.
Weekly GitHub question submissions are graded separately.
Discussion sessions meet weekly through Tuesday, November 17, 2026. Students may miss up to one discussion session without losing participation credit. Additional absences reduce this component.
To receive full participation credit for a session, students must:
- give a thumbs-up to at least one teammate during Group Discussion; and
- rate every presenting team and every challenger during Present and Challenge.
Students are encouraged to give a thumbs-up to every teammate who contributed.
Groups may be assigned randomly or formed voluntarily, depending on the activity. Any tools may be used during Group Discussion. During a presenting team’s turn in Present and Challenge, its members may not use notes, computers, AI tools, or assistance from anyone else.
If an in-class network or technical problem prevents a student from completing a required action, the student should tell the instructor during class and will receive full participation credit for that session.
Individual voting records will remain private and will never be released. The instructor may share aggregate results with the class, including the top-rated individuals or teams.
Submit questions through GitHub before class. Use Popping during Tuesday class.
- Login name: your NetID
- Password: the last four digits of your UIN
Tell the instructor during class if you cannot sign in.
Quizzes
There are four 30-minute, completely closed-book quizzes, given in class on Tuesdays before the discussion session. Quiz 1 contains 8 short-answer or multiple-choice questions because only two discussion sessions occur before it. Quizzes 2 through 4 contain 12 questions each. Together, the quizzes make up 25% of the course grade.
For each previous discussion represented on a quiz, three questions appear exactly as they were covered during that discussion and one comes from other student submissions. The latter may appear as submitted or with a slight modification. Quiz 1 therefore contains six discussion questions and two other submitted questions. Each later quiz contains nine discussion questions and three other submitted questions.
| Quiz | Date | Discussion topics covered |
|---|---|---|
| 1 | Tuesday, September 15 | Weeks 1-2 |
| 2 | Tuesday, October 6 | Weeks 3-5 |
| 3 | Tuesday, October 27 | Weeks 6-8 |
| 4 | Tuesday, November 17 | Weeks 9-11 |
Final Project
The final project makes up 25% of the course grade. Students work individually or in teams of no more than three. Each team selects its own data, defines a meaningful statistical question, and develops an analysis that includes statistical reasoning, computation, interpretation, and important limitations.
Using an AI coding agent for the final project is optional. An agent can make parts of the data analysis, coding, checking, and report workflow easier. If an agent is used, its contribution should be checked and documented clearly. One possible design is to build and audit a reusable skill that guides an AI agent through a justified, reproducible analysis and report. This is an example rather than a fixed project format.
Store the final report as project/final-report.pdf or a self-contained project/final-report.html; it must not exceed 20 pages. An appendix may be project/appendix.pdf or a self-contained project/appendix.html. Final presentations take place during Weeks 15 and 16. Milestones, deadlines, and detailed evaluation criteria are TBA.
Weekly Discussion Bonus
Each week, every member of the highest-rated presenting team and the highest-rated challenger will receive bonus points. Bonus points are not part of the 100% base grade. Point values, tie rules, and other details are TBA.
Letter Grades
Letter grades will be based on the final course percentage. The regular letter-grade thresholds and the A+ threshold are TBA.
Course Tools and AI Use
Students need:
- Git and a personal GitHub account;
- at least one AI coding agent, such as Codex, Claude Code, VS Code with an AI coding extension, or another coding agent the student prefers; and
- R or Python for statistical computing.
Students may use either R or Python unless an activity states otherwise. Both languages are supported in the lecture notes where practical. Week 1 provides instructions for making an AI coding agent available, setting up the course GitHub repository, and following the discussion-question workflow.
Students must have at least one AI coding agent available, but using it is optional unless a homework question explicitly asks them to test a skill. An agent can make course setup, GitHub work, coding, formatting, and checking easier. AI tools may not be used during quizzes. When students use AI assistance, they remain responsible for checking and explaining the work and for documenting meaningful assistance when requested.
See GitHub Setup and AI Tools, Discussion Questions: Procedures and Rules, and the university’s Generative AI Guidance for Students.
Academic Integrity
The University of Illinois academic-integrity policy applies to all course work. The relevant rules are stated in Article 1, Part 4 of the Student Code, including Section 1-402.
Collaboration is encouraged during designated group discussions. Each weekly GitHub question must represent the student’s own understanding and thinking. AI use is allowed for homework, but students are responsible for checking and understanding the completed work. Project work must accurately describe the contributions of team members and any AI tools used.
All quizzes are completely closed book and must be completed independently. Books, notes, course materials, websites, AI tools, and assistance from another person may not be used during a quiz. Using any of these resources during a quiz is an academic-integrity violation.
When the boundary is unclear, ask the instructor before submitting the work.
Changes to the Syllabus
The instructor may make changes that are academically advisable. Any changes will be announced in class and posted on this website. Students are responsible for checking course announcements.