Final Video Presentation Grading Rubric

This document outlines the requirements for the final video presentation for the Data Science Clinic at the University of Chicago.

The audience of this presentation is your peers. They are smart and knowledgeable about data science but are by no means experts in the field. The aim of this communication exercise is to create a record of what you did that is accessible, interesting and provides a high-level overview and summary of your work. It isn’t supposed to be complete or provide enough detail on how to replicate your work.

The following presentations are high quality examples from previous years. This should give you a sense of the level of effort that is expected (Note that some of the requirements have changed from previous years, so these would not be considered “A+” under this year’s rubric).

You can find example presentations here.

Requirements

• The video presentation should be between 8-12 minutes in length. This is a hard limit and the best presentations tend to clock in around 9-10 minutes.
• The video must be uploaded to Canvas. No links to YouTube, Google Drive, Dropbox, etc.
  • Canvas submissions are limited to < 75MB. Reduce the quality of the video to reduce the file size. Leave time to upload to Canvas before the deadline. Failing to meet the deadline or not uploading due to file size problems will negatively impact your grade.
• The presentation needs to convey:
  • The purpose and importance of the project.
  • The contribution that you made.
  • The results you have completed.

Logistics and Timing

There are two draft videos due before the final video. All video submission dates and times can be found on Canvas. For each draft video, you will receive feedback from your project mentor on what improvements or changes should be made. Those changes should be implemented before submitting the next version.

Grading

Your group will receive a separate grade for each draft and the final video. Each team member will receive the same grade. You will be graded on the following factors:

• Slide quality
  • Do the slides use correct grammar, consistent formatting/style, and are they easy to read?
  • Are the slides too busy?
  • Are plots/graphs labeled correctly and easy to read?
  • Are all images and graphics, including DSI and UChicago logos, in the appropriate aspect ratios (not stretched).
  • If you are using a DSI or UChicago logo in the slide it should not be blocked by text or graphics.
• Presentation quality
  • Does the presentation achieve the goals above (introducing the problem, your contribution and your results)?
  • Did the presenters speak loudly, slowly and clearly?
  • Are the speakers bringing energy?
• Organization
  • Is the presentation well-organized?
  • Does it follow an obvious pattern that the audience can follow?

Draft videos should demonstrate a significant level of effort. Each draft should satisfy the requirements listed above and avoid any placeholder slides/graphs/tables. The second and final drafts will be graded on if you responded to the feedback from your mentor.

First Draft Video (30 Points)

• 15 Points: Presentation
• 5 Points: Organization
• 10 Points: Slides

Second Draft Video (40 Points)

• 15 Points: Presentation
• 5 Points: Organization
• 10 Points: Slides
• 10 Points: Updates, changes and issue correction based on feedback

Final Video (60 Points)

• 20 Points: Presentation
• 20 Points: Organization
• 20 Points: Slides

Dos and Don’ts

• Do not speed up / slow down the presentation using software tools
• Presentations which are stitched together from multiple videos in obvious ways will be heavily penalized. It is distracting to the audience when things like resolution, screen focus, etc. change.
• Before submitting make sure that all plot axis are labelled and formatted properly for a high-level presentation. For example if you are using a percent turn it into a percent, rather than leaving as a decimal. Make sure that when presenting large numbers your plot doesn’t say “1e6” or the like at the top.
• The size of all axis labels and titles should all be large enough to read and understand.
• Make sure that plots have titles and that they are easy to read.
• You should not try to be comprehensive of your activities. The goal of this presentation is NOT to cover everything, but to provide a short introduction to and results of your work. In general shorter presentations are usually stronger.
• Use a microphone. The presentation audio needs to be well recorded. There should be no background noise, difficult to understand speakers, audio that goes in and out, etc.
• Avoid using terms which are specific to your project or technical jargon.
• The presentation should not contain screenshots of Jupyter Notebooks or other code.
• Do not attempt to teach advanced concepts. If a core part of your project is a novel or unfamiliar concept then you should explain, but at a high level.
• Make sure to spend time framing the problem, describing why it’s important and how what you are doing links to it.
• Every project has interesting and unexpected aspects – you want to highlight those. Are there any cool visualizations? What about results?
• Do not violate any NDA/share any special sauce.
• Do not read lists of numbers from the slides.
• Specific descriptions of technology tend to be low information and should be avoided. It’s okay to have a system diagram if the project has a large engineering component and the presentation of that system diagram is key to understanding the work. Simply stating “Here is the system we built” and walking through the diagram without additional context is not useful for this presentation.
• No meta-editorializing like “Hmmm.. Let’s fix this next time” or “ummm… I think we should change the order of the slides in the next draft”. Each video should be treated as a final version.
• Keep the mouse icon steady, do not block text or graphics or move it around the screen. It is distracting.
• Make sure that all presentation materials are visible in the window being used.

Late Policy

Assignments submitted after the due date will receive a grade of zero.

Best Practices

The following are a list of best practices that improve the quality of the video. Since these videos are shared via social media we strongly encourage following these practices if you have the necessary equipment.

• Use a real webcam, rather than your computer.
• Do not blur your background. If you wish to block out what is behind you, please use a background image.
• Frame yourself in the center of the web camera, rather than having your face be off to one side or another.
• Consider the lightning in your recording area:
  1. Do not have a window in the background as it will shadow your face.
  2. Use a bright light in front of you.
• When using Zoom:
  • Check “Record video during screen recording” to make sure your webcam video is included,
  • Check “Place video next to the shared screen in the recording” so as not to block the presentation
• Record your video in 1920x1080.
• We recommend using text animations to avoid having an entire slide of text show up at once.

F.A.Q.

Does everyone need to speak?
No. We strongly encourage everyone to participate and present at least a part of the final presentation, but we will not be assessing individual contributions.

How should we record the presentation?
Zoom, Canva (and even Microsoft PowerPoint) provide tools to be able to create a high-quality video. From our experience, Zoom has consistently been the easiest for groups to use, so that is our default recommendation.

What if my presentation video isn’t uploading to Canvas?
If you are struggling to upload your presentation, there are a few solutions that we have found to work: <ul><li>Make the video smaller (you can google different techniques for this depending on if you are on Mac or Windows).</li><li>You can try to zip your video.</li> <li>Make sure that you are on a high-quality internet connection.</li> <li>Try a different group member’s computer.</li></ul>

Grader Notes

The draft and second draft of the presentation are done by the faculty mentor while the final video is graded by the Director and Asst. Director of the Clinic. Students are required to upload the video to Canvas, so make sure to give feedback and grades in Canvas. The Rubric above which describes how many points should be awarded to each section should be followed. An example assessment can be found after this section.

The videos should be graded according to the following point assignment:

• Slide quality (10 Points)
  • Do the slides use correct grammar, consistent formatting/style, and are they easy to read?
  • Are the slides too busy?
  • Are plots/graphs labeled correctly and easy to read?
• Presentation quality (15 Points)
  • Does the presentation achieve the goals above (introducing the problem, your contribution and your results)?
  • Did the presenters speak loudly, slowly and clearly?
  • Are the speakers bringing energy?
• Organization (5 Points)
  • Is the presentation well-organized?
  • Does it follow an obvious pattern that the audience can follow?
• Feedback changes (10 points, second video only)
  • Did the students implement changes and correct issues based on feedback?

Remember that most of the grade for the presentation is weighted to the “final” deliverable. The goal for grading the drafts is to get the final deliverable into as strong a place as can be.

Generally the first video will need significant work, so make sure that the grading reflects that. It is not uncommon to give 50% or less of the points in a specific category if the students fail to meet the basic requirements (and specifically the “Dos” and “Don’t” section).

Things that are particularly irksome: if a student says “I’ll fix this next time” in the video recording or otherwise engages in meta-editorializing. This should be penalized harshly.

Do not be scared to be aggressive with your comments if there is behavior that you see which needs to be avoided. For example, if one of the presenters is being low energy or significantly worse than the other presenters, be clear with your feedback “Sam needs to bring more energy to their presentation” or “The presentation is uneven with some presenters being high energy and others low energy. Please find a level that all presenters can be consistent on.”

We require two drafts because, historically, they get much better each time. By the final video the quality is usually pretty good.

If you have any specific questions feel free to reach out by Slack.

Example Assessment

The following was an assessment provided to students on a specific project a few quarters ago. It is provided as a template.

Note that this was a good, but not great first effort.

---

Overall – very solid. There are a few overarching things that need to be changed and a few specific things that need to be updated. 12/15 Presentation, 4/5 Organization, 8/10 slides: 24/30.

• Remember your target audience! Your language goes deep and then shallow (in terms of complexity, use and expected level of data science sophistication).

• In a presentation of this length and with the intended audience you need to be intentional around the technical language you use. For example: you jump between “recall”, type I errors, type II errors language. This isn’t helping the average audience – pick a single language representation and stick with it.
• You should think of the following – every time you use special language it costs you a gold coin. In a presentation of this length, you get 3 coins (one every 3 minutes)
• “Neutrino Event Interactions” – what are these? You state them in the intro. What does this mean? Gold coin!
• For the model section, a similar set of changes you should consider:
  • CNN, U-Net, OOP, “run files”, “HDF5”, “CSV”, “weighted loss”, “PyTorch”, “overfitting”, “Percentile Model” all of this language appears in either your slides or verbal description, and all of it is technical and most of it can be avoided. These are all gold coins!
  • As an example, consider when you describe the data loader. Simply stating “We wrote software which allowed us to load images in parallel using a more efficient format preferred by Fermi, rather than other common images formats” <- is way more useful then ever saying “HDF5” and doesn’t cost a gold coin!
• The slide organization needs a bit of work, I got confused about the regression model (it’s not a contribution?) vs. the other models. At the start of the presentation when you lay things out you need to be more explicit with what you are doing so the audience can follow along.
• You repeated the explanation of the image colors on the slide from the 2nd to the 4th speaker. Better coordination! Is there a reason to show the same style of images twice?
• The format of the slides changed throughout the presentation – create a single style and everyone use it. Every time you switch styles it distracts the audience.