To start the new block, ‘Teaching’ with Data, I decided to collect my drinking habits throughout the week. For this block I will also try to transform the hand-drawn visualizations into a more “traditional” data analytics dashboard.
Throughout the week I recorded what I drank, whether or not I purchased the beverages from a restaurant/cafe, whether I also had a snack/meal with the beverage, and a non-numerical consumption amount (e.g. a few sips, a large tea, a glass of juice, etc).
I did not fully record what I drank on Thursday – the last entry recorded is the water I had with dinner. I noticed this error Friday morning and did not append the data since I had not recorded it on Thursday. This is an attempt at modeling tracking applications as many do not allow you to append the data yourself either. This hints to the fact that data sets are often representative and limited.
The data for the week resulted in the following visualizations.
Design and Discussion
The dashboard view was developed using Metabase (dependent on a local database [SQLite]) and Adobe XD. My hand-drawn visualizations are drawn using GoodNotes on my iPad.
In designing both visualizations, I found the most difficult aspect to convey was the amount consumed. Since I did not record a numerical value, I had to estimate one for each amount, which carries some assumptions: (1) I drank the entire bottled or purchased beverage, (2) I follow volume averages for “sips” and “gulps”, and (3) the initial volume was consistent across non-bottled beverages. For the hand-drawn visualization, I settled for the dashed line system above the drink category, where: single line is 4-9 fl oz, a double line is 10-15 fl oz, and a triple line is 16 and more fl oz.
For the dashboard view, I was limited not only by the lack of precise numerical values, but also the customization capabilities of the software and the translation of data into a database table. Consequently, I decided to use the number of dashes as the amount consumed. I also kept the system relatively simple – a single table with columns: drink (text), purchased (boolean), amount (integer), food (boolean), and day (text). In a more expansive system (and more robust database queries), some of the limitations, but not all, of the customization options could be overcome. However, in the classroom teachers may not have access to modify or view the structure of the backend. This restriction hinders a true and complete understanding of how the visualization is developed and what conclusions can be drawn.
Very stylish visualisation here Dillon.
‘However, in the classroom teachers may not have access to modify or view the structure of the backend. This restriction hinders a true and complete understanding of how the visualization is developed and what conclusions can be drawn.’
This is a useful conclusion, particularly given the very specific choices about representation that you’ve described here. This kind of reflection – ‘implications for teaching’ could be developed a bit here, and this would be something to try and elaborate upon for your next visualisation. Remember that the readings (in this case from block 2) are our primary source for identifying issues, ideas, and concerns for ‘teaching’ with data.
Hi Jeremy! I’ll be sure to include some of the literature in the visualization discussion this week!