Reading / Journal 12

Getting Visual ✏️

Journals are a combination of a few things. First, the "journal." This is a very small weekly assignment where you write 200+ words (option 1) or submit "tinker" code (option 2). Second, the "reading." This is where I direct you to read from a book, an online article, or watch a video, and I supplement those with my own lecture notes, graphs, figures, quotes, or so on.

All other assignments and lectures build upon these. Usually, there will be more reading at the start of a unit and less at the end.

Reading

In Labs 4 and 5 we visualize data.

A good visual will do two things, no more, no less:

• Show that there is a pattern in the data
• Show what that pattern is

Statistics can do the first, and qualitative methods ("looking at" the data) can do the second.

But a good visual–it can do both.

This visual represents the conversations students had in the first and second version of an educational game:

The red and blue dots are the individual students, red representing those playing the new version, blue the old version. The red and blue squares are the "means" of those two groups, and the dashed-line-box around the squares is the "confidence interval" of those means. Because the boxes don't overlap, we have good evidence that the two groups really did differ (that there is an effect from the updates of the game).

So what do the positions of the dots mean? What do the red and blue lines mean?

Those represent the connections students made as they talked to each other during game play. Students in the first game appear to connect Technical Constraints with Collaboration more frequently. In other words, they were asking each other, "what are we supposed to be doing?"

Students in the second game appear to connect Design Reasoning with Performance Parameters. In other words, they were doing the work of engineers, which the game was supposed to teach them.

That was a bit of a complicated visual, since what it is supposed to represent is complicated: how people talk about things.

This visual from Pew Research is simpler:

It shows how frequently different age groups, on average, report seeing content online that makes them feel a certain way.

I'm not convinced that there is a relationship between age and any particular feeling: The pattern appears to me to be that older people report seeing everything less frequently.

Next, consider this one:

I really like this one. As you may know, I research AI Ethics. But also, it shows the overall for/against a particular use of AI, while also showing why we see those differences. It gives us the big picture number and goes a little further into what's what: and good balance of Quant and Qual, which I always appreciate.

Last example, a violin plot:

This shows us that chicks fed on sunflower-based feed were, on average, heavier than those fed soybean-based feed. However, it also shows the distribution of those groups: We can see that some soybean-chickens were heavier than some sunflower-chickens, but we can still see trend is clear for most of the chicks.

No matter the particular visual you go with, it always comes down to units, groups, and measures. ("Units" is short for "unit of analysis," not "units" like feet/inches/pounds/etc.)

In the educational game example, the units are the individual students. They are grouped by which version of the game they played. The measures are the counts we tracked for how often they connected certain concepts.

In the first Pew example, the units are all the respondents to the survey. They are grouped by age. The measures are the frequency they report seeing content that evokes each emotion.

In the second Pew example, the units are again all the respondents. They are not grouped here. The measures are their yes/no answer to the question about a particular use of AI, and the category they picked for why they picked that yes/no. (You can actually see that some respondents appeared to pick the wrong yes/no on accident, based on the "why" answer they gave.)

And in the last example, the units are the individual chicks. They are grouped by what kind of feed we gave them. The measure is their weight.

The reason we pick different plots in those cases boils down to either: we are working with different types of data, like numerical, categorical, etc.; we want to show different types of relationships, like comparisons, flows, etc.; or both.

Now watch "Understanding GroupBy Operation" from https://www.youtube.com/watch?v=fujPIABgUD0

And read the "Grouping" section from "10 minutes to pandas" https://pandas.pydata.org/pandas-docs/stable/user_guide/10min.html

And familiarize yourself with "The Data Visualization Catalog" from https://datavizcatalogue.com/search.html

And read "Why scientists need to be better at data visualization" from https://knowablemagazine.org/article/mind/2019/science-data-visualization

And (optional) skim the first five "Data Visualization" tutorials from https://www.kaggle.com/learn/data-visualization

And read Chapter 16 from our textbook.

Journal

Choose from one of the following two options as you best see fit:

Option One: Submit a brief "200+ Words" reflecting on the reading and/or the course as a whole. These 200+ words are expected to come completely from the student, ignoring words from quotes/etc. The format of these assignments is up to the student as it best helps them: bulleted point notes on the reading; questions directed at the instructor of course material; a paragraph reflecting on the Lab assignment for the week; a poem; a summary of recent technology news; anything, so long as it is turned in on time, is relevant to the course, and meets the required length.

Option Two: Submit a brief "Tinker" where you have attempted to "program" something, using the tools of the course, that is not directly related to another course assignment. Include screenshots of the input work done and the output result (even if it does not work), along with a brief statement of your intentions, the approach you took in getting it to work, and your thoughts on your result so far. Make sure it is clear what code came from you and what came from online/the reading/etc.

Submission

Submit your journal as a Word/PDF (NOT a .pages) document to Blackboard.

Grading

Journals are each worth 1/100 towards your final grade. Grading is pass/fail based on meeting the requirements of the chosen option.