Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

Overview

Instructor: Scott Linderman
TA: Aymen Echarghaoui
Term: Spring 2026
Stanford University


Course Description

This course will teach you how to do applied statistics research. We will follow Box’s Loop: an iterative approach of asking a scientific question, collecting data to answer that question, building a model, performing statistical inference, and then criticizing and revising the model in light of your findings. We will develop the statistical tools to carry out this process: multivariate Gaussian models, graphical models, MCMC, variational Bayesian inference, latent variable models, state space models, Transformers, diffusion models, and more. We will practice this process through an extended research project and a flipped classroom with biweekly in-class lab meetings.

Prerequisites

Students should be comfortable with basic probability and statistics as well as multivariate calculus and linear algebra.

Logistics

We will alternate between traditional lectures on odd-numbered weeks and lab meetings on even-numbered weeks. The lab meetings will be a flipped classroom -- each project team will present their deliverable for that week. See the next section and the Course Project for more detail.

Assignments

Project Deliverables

There will be project deliverables due every two weeks on Sunday night at 11:59pm. We will have lab meetings the week following the deliverables in which each project team will briefly present their progress. Each team will be assigned to either the Monday or the Wednesday group, and they should only attend the lab meeting they are assigned to. See the Course Project page for more detail.

Math Problems

Additionally, there will be one math problem assigned each week, due the following Wednesday night at 11:59pm. These problems will help you test your reasoning abilities and, if you’re a PhD student, prepare for quals.

Schedule

The course will teach you the skills necessary to follow Box’s Loop: formulate a problem, collect data, build a model, perform inference, criticize, revise, repeat. Topics from Parts I–III (Foundations, Latent Variable Models, Inference Algorithms) are interleaved so that each new model is paired with the inference tools needed to fit it. Part IV (Sequence Models) occupies the final few weeks. We won’t cover Part V (Stochastic Processes) this quarter, except in passing, but some chapters will reference that material in case you want to dig deeper on your own.

As described above, the course alternates between traditional lectures and lab meetings, where we will flip the classroom. You will be assigned to either the Monday or the Wednesday lab meeting; you should not attend both. During lab meetings, you will give a short (1 slide, 3 minute) presentation of your deliverable, and you will give feedback to others.

Project deliverable due dates are marked below.

DateTopicReading
Mar 30Attend: Course overviewCh 1.1
Apr 1Attend: The (Multivariate) Normal DistributionCh 1.21.3
Apr 5Deliverable 1 Due (1pg report per person)
Apr 6Watch: Mixture Models
Attend: Lab Meeting (Monday Teams)
Ch 2.1
Apr 8Watch: Expectation Maximization
Attend: Lab Meeting (Wednesday Teams)
Ch 3.3
Apr 10Required: You must find your teammate by this date.
Apr 13Attend: Hierarchical ModelsCh 1.5
Apr 15Attend: Markov Chain Monte CarloCh 3.1
Apr 19Deliverable 2 Due (2pg report per team)
Apr 20Watch: Probabilistic PCA
Attend: Lab Meeting (Monday Teams)
Ch 2.3
Apr 22Watch: Variational Autoencoders
Attend: Lab Meeting (Wednesday Teams)
Ch 2.4
Apr 27Attend: Hidden Markov ModelsCh 4.1
Apr 29Attend: Linear Dynamical SystemsCh 4.2
May 3Deliverable 3 Due (2pg report per team)
May 4Watch: Switching Linear Dynamical Systems
Attend: Lab Meeting (Monday Teams)
Ch 4.3
May 6Watch: Recurrent Neural Networks
Attend: Lab Meeting (Wednesday Teams)
Ch 4.4
May 11Attend: TransformersCh 4.5
May 13Attend: Deep SSMs and Linear AttentionCh 4.6
May 17Deliverable 4 Due (2pg report per team)
May 18Watch: Parallelizing Nonlinear RNNs
Attend: Lab Meeting (Monday Teams)
Ch 4.7
May 20Watch: Continual Learning
Attend: Lab Meeting (Wednesday Teams)
Ch 3.7
May 25No class — Memorial Day
May 27Attend: Diffusion Models and SDEsCh 2.5, 5.2
Jun 1Attend: Project Presentations (All Teams)
Jun 3Attend: Project Presentations (All Teams)
Jun 8Final Report Due

Grading

ComponentWeight
Milestones (4 × 10%)40%
Final report35%
Lab meeting participation15%
Weekly math problems10%

Each milestone is graded on an A-F scale, roughly following this rubric:

ScoreMeaning
AComplete, thoughtful, and well-executed
BAcceptable but missing key elements or depth
CDid not take assignment seriously / just asked AI to do it
FNot submitted or substantially incomplete

We may assign half-letter grades too.

Math problems will be graded on a (0, 1, 2) scale.

Remember that project grades are based on the quality and thoroughness of your applied statistics practice — not on whether your model achieves impressive results. A project that honestly finds that a simple baseline outperforms a complex model, with a clear explanation of why, is an excellent project.

Books

In addition to the lecture notes, you may find these textbooks helpful: