Advanced Statistical Modeling
Spring 2021, Stanford University
Spring 2021, Stanford University
Introduction to high-dimensional data analysis and machine learning methods for use in the behavioral and neurosciences, including: supervised methods such as linear regression and classification, linear mixed-effects and hierarchical models, structural equation modeling, and regularization techniques; statistical methods such as bootstrapping, signal detection, and reliability theory; metrics for model/data comparison such as representational similarity analysis; and unsupervised methods such as clustering, PCA, and exploratory factor analysis. Students will learn how to both use existing statistical data analysis packages (such as scikit-learn) as well to build and estimate simple custom models in Python. Requirement: Psych 251, and familiarity with Python programming and introductory linear algebra.
Time: Mondays & Wednesdays 13:00 - 14:20 PDT.
Location: Virtual! (Email or Slack instructors for Zoom link.)
Staff:
Daniel Yamins (x@stanford.edu where x=yamins)
Russ Poldrack (x=russpold)
Chengxu Zhuang (x=chengxuz)
Course goals:
Students completing Psych 253 will:
Prerequisites: Familiarity with basic data handling and statistical methods (e.g. Psych 251 and Psych 252, or similar); intermediate-level fluency in Python programming; and comfort with introductory linear algebra (e.g. Math 51, EE103, CME100, CME 104 or similar).
Homeworks: There will be two homeworks assigned throughout the quarter. For each assignment, students will submit a lab report containing their results. Lab reports must be in the form a runnable Jupyter Notebook, together with a PDF converted from the notebook.
Final Projects: Final projects will consist of an application of techniques learned in the class to students' own datasets (for students without datsets, we will supply data from several recent neuroscience experiments). Student will present their results to the class, and subsequently submit a Lab Report describing the methods used and results obtained.
Grading basis: Lab reports (40%), final project (30%), class participation (30%)
Office hours: Tuesdays 10:30-11:30 am PDT (email or slack Dawn for Zoom link).
Github Repo: https://github.com/neuroailab/psych253
Slack: stanfordpsych253.slack.com
Location: Virtual! (Email or Slack instructors for Zoom link.)
Staff:
Daniel Yamins (x@stanford.edu where x=yamins)
Russ Poldrack (x=russpold)
Chengxu Zhuang (x=chengxuz)
Course goals:
Students completing Psych 253 will:
- be able to apply existing supervised or unsupervised analysis and machine learning methods to experimental data using Python packages including Numerical Python (Numpy) and Scikit-Learn.
- understand how to implement key existing methods from scratch in Python, and develop their own simple custom analysis methods
- have mastered best practices for experimental data analysis and visualization, including comfort with the HDF5 data format, IPython notebook and Matplotlib.
Prerequisites: Familiarity with basic data handling and statistical methods (e.g. Psych 251 and Psych 252, or similar); intermediate-level fluency in Python programming; and comfort with introductory linear algebra (e.g. Math 51, EE103, CME100, CME 104 or similar).
Homeworks: There will be two homeworks assigned throughout the quarter. For each assignment, students will submit a lab report containing their results. Lab reports must be in the form a runnable Jupyter Notebook, together with a PDF converted from the notebook.
Final Projects: Final projects will consist of an application of techniques learned in the class to students' own datasets (for students without datsets, we will supply data from several recent neuroscience experiments). Student will present their results to the class, and subsequently submit a Lab Report describing the methods used and results obtained.
Grading basis: Lab reports (40%), final project (30%), class participation (30%)
Office hours: Tuesdays 10:30-11:30 am PDT (email or slack Dawn for Zoom link).
Github Repo: https://github.com/neuroailab/psych253
Slack: stanfordpsych253.slack.com