Course:FRE521E
| FRE 521E:
Economic Analysis Using Machine Learning | |
|---|---|
 | |
| FRE 521E | |
| Section: | |
| Instructor: | Kevin Laughren |
| Email: | kevin_laughren@sfu.ca |
| Office: | |
| Office Hours: | TBD |
| Class Schedule: | Feb 23 to April 10
Tuesday 10:30- 12:00 |
| Classroom: | MCML 154 |
| Important Course Pages | |
| Syllabus | |
| Lecture Notes | |
| Assignments | |
| Course Discussion | |
| [[Category:]] | |
COURSE DESCRIPTION
This course introduces modern machine learning methods in R through the lens of applied economics, with a focus on climate, food, environmental, and resource applications. Students learn why and when machine learning can outperform traditional econometric approaches—particularly for prediction-focused problems, high-dimensional datasets, and cases where relationships are too complex for standard functional forms.
Through assignments and a final project, students will have opportunities to code, experiment with the machine learning models, and communicate their modelling choices and findings in clear, accessible language best suited for non-technical audience.
Real-world applications include:
- Using satellite imagery and high-dimensional data to predict crop yields and classify land cover
- Using spatial and sensor data to predict environmental outcomes such as air quality
- Using large government and customs datasets to predict trade volumes and trade flows
LEARNING OUTCOMES
By the end of this course, students will be able to:
- Identify when an applied economics problem could benefit from a machine learning approach (e.g., prediction problems, large amounts of unstructured data, modelling high-dimensional relationships without restrictive functional forms)
- Describe the trade-offs associated with machine learning approaches versus classic econometric approaches
- Apply modern machine learning methods such as random forests, boosted models, and neural networks to generate economic insight
- Communicate economic insights using the outputs of machine learning models.
ASSESSMENT REPORT
To be Updated Soon
REAL-WORLD APPLICATIONS IN CLIMATE, FOOD & ENVIRONMENT
- How can machine learning be used to predict environmental and agricultural outcomes using high-dimensional data, enabling more efficient investment, planning, and risk management?
- How can machine learning help us make better predictions in climate, food, and environmental systems—especially when working with high-dimensional, complex, or unstructured data?
- How can we create smarter, more cost-effective data collection strategies—such as adaptive sampling—to reduce data costs while improving the accuracy of environmental, agricultural, or inspection programs?
ASSESSMENT METHODS
To be Updated soon
| Learning Assessment Activity | Percentage | Dates |
| Participation | 10% | Weekly |
| Assignments (best three) | 45% | End of Weeks 2, 3, 4, and 6 |
| Project Outline | 10% | End of Week 5 |
| Final Project | 35% | End of Week 7 |
| Total | 100% |
ASSESSMENT METHODS
All assignments, exercises, and your project are required to include both your code (in .r or .rmd format) and a write-up (in .html or .pdf format) that is written and formatted to a professional standard. For assignments where you do not use data provided by the instructor, you will also need to submit a data file (in .csv format). The code should be written so that it loads the relevant data file and compiles on an external computer. You are encouraged to have a peer test that your code compiles on their computer before submitting.
Unexcused late assignments are penalized 25% of the grade weight immediately after the deadline, and an additional 25% penalty applies for additional day an assignment is late. For example, assignments that are fifteen minutes late lose 25%, assignments that are one day and fifteen minutes late lose 50%.