Data Science 2 with Python (303-2-20)

Instructors

Lizhen Shi

Meeting Info

2001 Sheridan GPC 2447: Mon, Wed 12:30PM - 1:50PM

Overview of class

Only Statistics majors, Data Science majors, Data Science minors, and Applied Statistics Masters students assigned to take 303-2 in this quarter are able to register for this course.

STAT 303-2 is the first course in predictive modeling in the STAT 303 sequence. It focuses on linear models and foundational techniques in supervised learning, emphasizing both statistical understanding and practical implementation in Python. Topics include linear regression, logistic regression, model assessment and validation, cross-validation, regularization methods, and feature selection. Students will learn how to build, evaluate, and interpret predictive models, understand the tradeoffs between model complexity and generalization, and apply these tools to real-world datasets.

Registration Requirements

Learning Objectives

At the end of the course, students should be able to:
Formulate predictive modeling problems (regression vs. classification) and prepare data appropriately for modeling.
Fit and interpret linear models, including linear regression and logistic regression, and understand key modeling assumptions and limitations.
Evaluate and compare models using appropriate performance metrics and validation strategies such as train/test splits and cross-validation.
Control model complexity using regularization and feature selection, and explain the bias-variance tradeoff in practice.
Communicate results clearly, including modeling choices, interpretation, and performance, using written summaries and reproducible Python workflows.

Teaching Method

Each class meeting combines pre-class reading, an interactive lecture with Poll Everywhere concept checks, and a brief end-of-class quiz (most days).

Evaluation Method

Small assignments (10%): brief end-of-class quizzes in most lectures.
Assignments (24%): five coding assignments, submitted as Quarto-rendered HTML; the lowest score is dropped.
Midterm exam (20%): in person, paper-based.
Final exam (25%): in person, paper-based, during finals week.
Prediction problem (15%): a Kaggle-style binary classification challenge with several checkpoints, graded on meeting performance thresholds rather than ranking.
Attendance (6%): recorded through Poll Everywhere concept checks.

Class Materials (Required)

An Introduction to Statistical Learning with Applications in R' by James, Witten, Hastie, Tibshirani (2013), with Python codes https://github.com/JWarmenhoven/ISLR-python, ISBN-13: 978-1461471370 (available for free online)

Class Materials (Suggested)

Linear Models with Python by Julian J. Faraway
Python data science handbook by Jake VanderPlas
The Elements of Statistical Learning by Trevor Hastie, Robert Tibshirani and Jerome Friedman

Class Attributes

SoC Mathematics, Science, and Technology Distro

Enrollment Requirements

Enrollment Requirements: Prerequisite: STAT 303-1
Add Consent: Department Consent Required