ECON 111 Machine Learning in Economics: Prediction and Fairness

Course Overview

Today, machine learning models are increasingly used to guide high-stakes economic decisions. Firms use data-driven systems to allocate scarce resources, from screening job candidates to deciding who to approve for a mortgage. These systems promise speed, scale, and standardized decision rules, but they also raise difficult questions: when the model approves someone who turns out to be a poor match or rejects someone who would have succeeded, do these errors fall unevenly across groups? Can algorithms unintentionally reproduce historical inequalities embedded in the data they learn from? And what should we consider a fair prediction system?

In this course, you will build the technical skills needed to investigate these questions by working with data and prediction models every day in class. You will learn to explore and visualize data, build all kinds of predictive models, and critically evaluate how those models perform and how their predictions would affect economic opportunities across different groups.

Schedule

Grading

Attendance

Because most of the work in this course is completed in class with your group, regular attendance is essential. If you expect to miss a significant number of classes, this course is unlikely to be a good fit.

You may miss up to two classes without penalty. Each additional absence will reduce your final course grade by 5 percentage points. Being more than 10 minutes late will count as an absence. Not being physically present counts as an absence.

If an unexpected situation arises, such as a major medical emergency, notify your groupmates as soon as possible. In that case, instead of completing the classwork during class, you will meet with your group to complete the classwork over Zoom during your TA’s office hours.

Labs will generally be for taking (and retaking) quizzes. If you are happy with your quiz grade after the first attempt, you are free to skip lab on the retake. I’ll keep the higher of your score on the first attempt and the retake for each quiz.

Classwork and Homework: 20%

During each class meeting, you will spend the entire period working on a classwork assignment in a small group. These assignments are designed to build your skills and help you apply machine learning methods to solve economics-related problems. Classworks will be due at the end of the class period (one copy per group, uploaded to Canvas). Homework will be due before the next class period (to be completed and turned in individually).

You will work with three different group sets over the quarter:

• Weeks 1–3: Groups will be randomly assigned.
• Weeks 4–6: Groups will be randomized again.
• Weeks 7–10: You may request groupmates for the final group set. I will try to accommodate as many requests as possible.

Quizzes: 80%

There will be five quizzes during the quarter, taken during your Friday lab, starting in Week 2. Each quiz has a retake opportunity the following week. The retake will cover the same material but will be focused on working with a different data set. I will keep the higher of your two scores from the original attempt and the retake. If you are happy with your first score, you are free to skip lab during the retake week.

To study, review the classwork assignments, since the quizzes are closely based on that material. Because quizzes make up 80% of your final grade and there are five quizzes, each quiz is worth 16% of your final grade. There will be no additional makeup opportunities, since the built-in retake schedule will already give you a second chance.

Grading Scale

At the end of the quarter, I will take your final grade on Canvas and round to the nearest integer (an 89.49 becomes an 89 and an 89.50 becomes a 90). Then I’ll apply this grading scale:

• A+: 97 or higher
• A: 94 - 96
• A-: 90 - 93
• B+: 87 - 89
• B: 84 - 86
• B-: 80 - 83
• C+: 77 - 79
• C: 74 - 76
• C-: 70 - 73
• D+: 67 - 69
• D: 64 - 66
• D-: 60 - 63
• F: less than 60