CMSE381 - Honors Project Addendum

CMSE381 - Honors Project Addendum#

If you are taking CMSE381 for honors credit, you will also complete a project, but you have additional requirements.

Getting Honors Option Set Up#

Please note that prior to doing the honors version, you need to discuss and get approval of the data set from Dr. Zhang or Dr. Bao. Make sure that you have been granted access to the #cmse381-ss25-honors channel for notes and announcements.

In addition, you are responsible for getting the university paperwork requirements for getting this set up, i.e. submitting the honors option form. Here is a template for the description in the option form:

“This honors option requires me to complete additional analysis for the semester project. The regular semester project require using a classification method and an additional method to construct machine learning models on a complex dataset. The honors option requires me to perform one additional method for each of the two methods and compare their performance quantitatively and qualitative. The deliverables for this project are: source files for any relevant code, a jupyter notebook where the data analysis and/or computational modeling is performed, a video presentation in which I present my findings, a PDF of the slides I used in the presentation, and any other relevant files. The project topic will be discussed with the instructor and finalized after the middle of the semester. I understand that in order to earn the honors option credit, I must receive an overall grade of 3.0 or higher and an honors project grade of 80% or higher.”

Project requirements#

In place of the CMSE381 project, you will complete a project which is similar, but with additional requirements. Here, I will include notes on what modifications you will submit from the standard project.

  • You will use the default dataset of your section or propose your desirable one(s), which needs to be approved by your section instructor. Appropriate data sets will be generally larger in scope, and are likely more messy requiring more data cleaning.

  • For each method (classification + one other method) you used according to standard project description, you must use an additional method and compare the modeling results with the first method (the type has to match: classification compared with classification, regression compared with regression). The two methods in comparison have to be sufficiently distinct (e.g., cannot be one linear regression model with vs without interaction term). You should discuss with your section instructor as you develop your project.

  • All four of these models must include discussions of general improvements and parameter selection such as k-fold CV or subset selection.

  • Your analysis should include an in depth discussion of interpretation of results, limitations of your work, as well as what you might do in the case of a followup project. This is the kind of thing that would go in the discussion section of a published research paper.

  • Overall, I am looking for your submitted notebook to be as similar in style to a research paper as possible. It should have an introduction with broader context of the data. It should be clearly annotated and discussed for me to understand what you are doing.

  • This can be a group project. Just make sure you clearly state your contribution of the two additional methods and relevant comparisons in the jupyter notebook.

  • Please note that as per CMSE policy, students must earn a 3.0 or greater in a course in order to be eligible for Honors Option credit.