Principles of Causal Inference: Term Projects
|
Students enrolled in the course are required to complete a small research project on causal inference on a topic to be chosen in consultation with the instructor. A written report on the project and a brief oral presentation summarizing the same is expected at the end of the semester. You may choose to work individually or in small groups (of at most 2 members) on the project. An ideal project aims to answer some interesting research question(s) related to the foundations, methods or applications of causal inference, or offers a critical, insightful, review of literature on a key topic (identifying key problems, organizing existing methods into thematic groups, identifying major gaps, etc.), or introducing advances in causal inference to an important new audience which could benefit from it (e.g., behavioral science, political science). The results of the project should be written up in the form of a scholarly manuscript modeled after articles that appear in an appropriate peer-reviewed journal or conference proceedings
You may make use of all the resources available at your disposal, including the published work of others, publicly available code, publicly available data sets, as well as consultation with others (fellow students, faculty, or other experts on the topic of your project). Note however, that your use of these resources should be guided by the best practices of academic research and writing.
In particular, you should exercise utmost care to avoid plagiarism: the deliberate use of someone else's language, ideas, data, code, or other original material that is not common knowledge without properly acknowledging the source. You should also familiarize yourself with appropriate ways to acknowledge the contributions of others and to cite all your sources (See for example, resources for avoiding plagiarism: http://www.plagiarism.org).
When you work in a team, each team member is expected to contribute to all aspects of the project, including conception of the initial idea, planning, implementation, experimental evaluation, and organization and writing of the report, and presentation of the results. However, because each individual may bring unique abilities to a team, and one of the goals of working in a team is to take advantage of the unique abilities of the team members, it is not unusual for the contributions of individual team members to vary across tasks. To ensure that each team member gets credit for his or her contributions, the final report should include a statement of contributions that explicitly identifies the contributions of each team member and a statement that each team member concurs with the contents of the report.
To avoid possible misunderstanding, it is advisable to for the each group to meet with the instructor and discuss each member's role with the instructor before beginning to work on the project.
If there are irreconcilable differences among members your team, you should notify the course staff as early as possible (but after having made a good faith effort to resolve the differences among yourselves) so we can help resolve the differences or suggest alternatives.
In the event that the dispute among team members is not resolved to everyone's satisfaction, the instructor's assessment (if necessary, based on discussions with by each member) will be binding.
If your background is in another discipline (e.g., engineering, life sciences, health sciences, behavioral sciences, cognitive sciences, business, social sciences, political science, etc.) you are encouraged to pick a
topic that would allow you to explore the development and/or applications of causal inference to a address research questions of interest in that area
Suggested Topics
The list of topics given below is meant to be suggestive, but not
exhaustive.
- Original analysis of existing data (e.g., related to public policy, healthcare, social science, etc.) through the causal lens (under alternative causal assumptions) to answer specific research questions
- Rigorous empirical comparison of existing methods for estimating causal effects from observational data, learning causal models from observational data, etc.
- Development of improved methods for estimating causal effects or learning causal models from observational data
- Extensions of existing methods to new settings, e.g., longitudinal data, network data
- Novel applications of methods for causal transportability or meta analysis.
The descriptions given here are rather brief, so please feel free to talk to the instructor to explore
possible project ideas in greater detail. You are also encouraged to look at current research in
the Artificial
Intelligence Research Laboratory for project ideas.
Term Project Timeline
- Form your group and select a project or term
paper topic. Prepare a brief (1 paragraph) outline of the project
or term paper and list of group members as well as the role of
each member is due on or before March 30, 2023.
- Write up a brief plan of work, including the aims
of the project, the approach to be followed, and anticipated results along with the relevant bibliography. April 6, 2023.
- An initial draft of your paper (in PDF form), data, code (if any), results, etc. as appropriate. April 20, 2023
- The final version of the paper accompanied
by all the source code (python or R notebooks), data, and experimental results, as appropriate. May 2, 2023.
Instructions for Preparing the Term Paper
The project will be graded on the basis of originality, technical soundness, organization, clarity of presentation, grammar and style, adequacy of the bibliography, as well as the significance of the results (as appropriate for the nature of the project). In short, it will be evaluated as though it is being refereed for publication in a journal or conference proceedings. You are encouraged to look at papers published in one of the major national or international conferences (e.g., AAAI, ICML, NeurIPS) or appropriate disciplinary journals that publish articles on methods or applications of causal inference as a model for your term paper. You are
strongly encouraged to use Overleaf (which offers templates for many journals and conferences) to prepare your paper.
Additional resources on research and writing can be found on the Graduate Research
and Writing page.
|
