BeanVis: An Artistic Visualization of Caffeine Intake

Abstract: As part of a course on personal informatics design, I visualized my daily caffeine intake over 44 days using a bar chart made entirely out of coffee beans (which I call BeanVis). The process of making this chart got so tedious and nettlesome that it gave me a new perspective on my caffeine intake. I grew hyper-aware of how much I drink on a daily basis and feel obligated to lower that intake.

Project Background

I started tracking my caffeine intake in late September and still do so. I have known for a while that my body metabolizes caffeine differently than the average person — I am highly sensitive to this molecule. I wanted to get a better idea of how caffeine might affect other patterns in my life, especially anxiety, irritability, and sleep, which I also track.

My project visualizes a 44-day sample of my caffeine consumption (from October 1st, 2016 — November 13th, 2016), using a bar graph made entirely out of coffee beans (635 beans to be exact). Consumption was measured using a custom scale consisting of “caffeine units”, described below.

Caffeine Units

The caffeine content of consumables is measured in milligrams. Most beverages have a consistent amount of caffeine, shown in the figures at the end of this report. My main sources of caffeine are coffee, tea, and diet soda. Over time, I have calibrated and titrated my intake per my own personal index of caffeine’s effect on my physiology. My unit of measurement is based on this personal index, resulting in “caffeine units” or “cu”.

1 cu equals an average cup of coffee, which is shown to have roughly 120 mg of caffeine. Hence, a cup of black tea equals ~0.5 cu, and a can of diet soda equals ~0.3 cu. Using the cu-scale, I have recorded my caffeine intake to a single decimal point, creating a measure that has functional meaning for me. The total caffeine units consumed during the 44-day period was 63.5, requiring 635 beans to represent the data.

Caffeine consumption data


The first observation is about the data itself: the chart reveals no consistency to my caffeine intake pattern. There is no correlation with weekdays vs. weekends, nor stressful time periods. This leaves me still wondering what influences my caffeine intake levels.

The second observation is about the process of creating BeanVis, which was extremely tedious. Placing 635 coffee beans into perfect rows grinds on you (pun intended). At some point, I started thinking to myself “I wish I drank less coffee so I wouldn’t have to place any more damn beans!” I saw this as an insight about how crafting a non-digital visualization can shape our perception of the things we’re tracking, and help augment our behavior change goals. In other words: personal informatics isn’t simply about generating data, it’s about connecting with it in a meaningful way.

(1) Creating BeanVis; (2) All the beans that were used for this project
Source: Randy Krum, 2010 (Image Available on
Caffeine content of tea by type (Source: Sophia Uliano; 5 Teas With Extraordinary Health Benefits)