This assignment does not count toward the final grade.
6 Oct 2021 by
a text entry box or a file upload
For more information, see slides and recording about the project and project proposal.
1.Find a project partner (or decide to do alone)
- Sign up to a group in Canvas or email me if you don't have Canvas access. If you don't have access but have a project partner who does, they can hand in the assignment.
2.Identify a causal inference problem (e.g., connected to a hobby, from your research, related to a MSc thesis, etc)
- It can relate to, for example, causal effects, decision making & policy evaluation, counterfactuals, understanding causal structure or all of the above.
- You are not meant to solve the problem in this proposal. Your goal is to identify a problem with a clear causal component that you think could be addressed during the course.
- Describe your problem in as much detail as you can. There are example problems on the slides from Lecture 1: Follow-up (and in the Course introduction). Examples include:
- Estimating the effect of an intervention
- Modelling counterfactuals
- Learning sequential decision-making policies
- Discovering causal structure
- Studying mediation effects
- Write up to 1/2 page describing the problem
3.Propose a dataset to learn from / experiment to run + identification strategy
- How will you identify the causal quantity of interest? If you are evaluating decisions, do you have data on those decisions? What led up to them?
- Will you be running an experiment? If so, how? What is the target of the experiment?
- If you learn from observational data, how will you identify the causal parameter?
- Write up to 1/2 page describing your identification strategy and dataset.
Grading & feedback:
- The proposal itself is not graded but the final project will be graded from 0–20 points. For a passing course grade, at least 5 points are required. See more details on grading on the course home page.
- I will give feedback on each proposal either in Canvas or by email.
Example projects from previous year:
Causal discovery in bike sharing service database [Observational study]
Improving prediction using causal graphs [Experiment / observational study]
- Causes that affect the performance of neural networks on the MNIST dataset [Experiment]
Causal effects of natural language using Yelp reviews (what in a review texts causes a certain review score?) [Observational study]
Causal effects of multiple concurrent treatments using the MIMIC database [Observational study]
Counterfactuals in Alzheimer’s disease and Lung cancer [Observational study / Simulator]
Impact of demographic variables on voting [Observational study]
Upload the proposal as a solution to this assignment (as a .pdf) before the deadline. If you are working alone or don't have access to Canvas, please email me.
You've already rated students with this rubric. Any major changes could affect their assessment results.