Caffeine and Mood Data Exploration

First thing in the morning, I head to a quiet conference room. I always bring the energy - and my thermos. Flying through the agenda, I lead the meeting like a caffeine-fueled cheerleader. Hours later, between sips of cold brew, I endlessly chatter during a class-wide discussion on experimental film. Walking to the library after class, I feel what I can only describe as an overwhelming sense of inspiration and joy. As I settle into my seat, my wild ideas manifest into jitters and I’m unable to focus. “What’s going on with me?” I wonder as I struggle to get to sleep at night.

Caffeine. It runs my college life. I consume it prescriptively and strategically drink it to improve my mood and productivity on the daily. But with a spike of energy being followed by a day-ruining crash, I wondered if there was a distinguishable correlation between coffee consumption and how I felt. As I began to talk to my friends, it seemed like it impacted them differently. “Like any drug it enhances my current state, so around my friends I’m in a happy mood; it makes me funny but alone and I’m stressed, it sends me into anxiety town”. With 5 coffee shops on campus, each offering a different menu of options, everyone has a method to the madness. One friend mentioned, “Caffeine from tea helps my brain, but caffeine from coffee makes me feel bad”. Inspired by mood diaries, me and 3 classmates began to collect data.

The Dataset

The data collection method was designed to track how caffeine habits cause mood fluctuations throughout the day. Participants kept a mood journal, taking note of mood and location every 2 hours as well as the times when they consumed caffeine and how much they slept the night before. The following is a preview of the dataset, and the rest can be viewed here.

Chat GPT Data Cleaning Exploration

Rather than have a limited set of moods participants could choose from to define their mood, there were a number of guiding adjectives given to generate creative responses. The result of this was a lot of moods without a clear scale to define their interaction. In search for a scale, this metric emerged: In order to standardize the emotions in the dataset, I defined a few of the most common responses in accordance with this spectrum, then fed ChatGPT a prompt to quickly standardize the rest. GPT also provided me with a python script which I used to populate a column of mapped emotion values and format the dataset. The resulting and final dataset looked like this can be viewed fully here.

Early Data Explorations

Using tools including tableau and Flourish, I looked for trends within my data to form a narrative surrounding it.

Our moods are all over the place, filling every quadrant of the grid, with the majority however, being the positive in the upper right hand corner. Cathy expressed being tired more often and consumed caffeine the least amount of times.

Exploring Difference by Person First

I noticed differences in the caffeine habits of me and my classmates. One classmate tended to drink large amounts of caffeine first thing in the morning and again in the afternoon. I tended to do so throughout the day, and another classmate had it only during the middle of the day.

This graph illuminates the difference in when me and my classmates drink caffeine, and how much we drink. It also highlights that my moods are stronger, and that Srishti drinks large amounts of caffeinated beverages.

Exploring My Moods and Caffeine Consumption

I chose to shift to an individual level, focusing on how my awakeness levels were altered following an increase in caffeine. I think the second day well illustrated a caffeine crash. This was a day where I had only slept 5.5 hours the night before and I drank a lot of caffeine in the early afternoon to mitigate tiredness. However the lack of sleep caught up to me and my awakeness began to decline.

Physical Product

The final iteration of this project required a physical manifestation, so I chose the vessel that houses the coffee itself. Some people in my class highlighted the way that water bottles have found a place beyond simply being a source of hydration, explaining they can also be an attachment object and that status and personality are communicated with their appearance. Additionally, I thought about how if you see someone drinking out of a thermos versus a water bottle, you expect they’ve got some hot, usually caffeinated beverage in there. What you don’t see, however, is what that drink is doing to them. I wanted to combine these ideas about communicating personality and effect of caffeine through a data visualization on my coffee cup. So, rather than simply printing a static data visualization onto my cup, I made it interactive.

Gaining popularity in asmr videos, pop-its emerged on the market as a fidget toy. With the material being readily available and commonplace to my gen-z peers, I cut and pasted the convex dots onto the outside of my cup. Maintaining the language of the prior visualizations, moods are encoded by color, with the placement of each dot symbolizing a time when caffeine was consumed. Following this placement method, a cup could look wildly different based on the individual’s internal experience. By adding the interactive element, this product becomes a performance piece, a dance depicting the jitters.