Today, the search for protein-rich foods is a staple in fitness and muscle-building diets. But what else are we consuming alongside these proteins? The following charts can help us make a more holistic decision about our food choices by measuring nutritional qualities beyond protein – such as fat and fiber – along with their environmental production costs – namely carbon footprint, land usage, and water consumption.

The charts were created using data from the US Department of Agriculture for nutritional values and from Our World in Data for environmental information. About 50 types of food were classified into three main groups based on their source of protein: animal, dairy, and plant-based. The selection focused on pure ingredients in their raw form, prioritizing higher protein values while also considering notable foods like tofu and mushrooms.

The aim was to find tangible ways to express the several qualities of each food and compare them, adopting units and portions familiar to an average person. To achieve this goal, a dataset and a series of charts were created. The dataset joined the data from both sources, and several more variables were calculated to explore ways of equalization of the foods. And three groups of charts were developed in Tableau: radar charts, scatter plots, and bar charts. For each group, the different food variables were intersected, allowing users to access multiple approaches and parameters for a more elucidative comparison.

Radar Charts

A radar chart model presented a convenient way to broadly cover both the nutritional qualities and the environmental production costs. With eight axes equalized for a 100g portion, several variables can be showcased in different scales of measure and proportions. However, the values with the same unit of measure, such as grams, were not modified.

On the right side of the circle were pulled the nutritious qualities of fiber, calories, fat (lipid), and cholesterol (at the bottom). On the left side are the environmental costs of the carbon footprint, use of land for production, and water withdrawn in the process. Lastly, the protein, chosen to conduct the study, was displayed at the top of the radar.

In order to highlight patterns among the three categories of foods, the solution was to unfold them into three extra radars in addition to the one with all three categories overlaid. This also allowed it to include information about the subcategories, such as red or white meat, and present more distinguishable resulting polygons. The radars were produced with an extension called LaData Viz, which is offered by the Tableau desktop software.

Next steps
At that moment of the creation process, the overall shape and general guides were defined. However, there are steps ahead to improve the chart. The first one would be finding a way to indicate the values for each radar in each of the variables. Possible tests would be a list aside from the radar or, in the case of interactive visualization, contextual labels that can show all the interesting data by hovering the mouse over each polygon.

A feedback was received about the color chosen for the three main categories, which might show bias against the animal source of protein. The pink being too close to the red could express negativity in this case. It will be valuable to verify such impressions with other users. Also, regarding the colors, it might be worth more tests to distinguish the sub-category’s tonalities while maintaining proximity with the main one.

Scatter Plots

Another way to demonstrate the relationships among various attributes of food was to print scatter plots. This type of chart enables it to incorporate colors and sizes to the points along with the two axes variables. The solution was to keep the protein attribute and the categories consistently on the x-axis and color, respectively, among the three chart variations while switching the other two dimensions (y-axis and size).

The first chart of the series adopted the carbon footprint as the y-axis, while the plot diameter represented the number of servings within a 100g portion. This distribution clearly shows the differences in portion sizes among the three categories, with the animal source presenting significantly smaller plots than the other two categories. This suggests that a holistic decision may consider multiple related factors. A typical portion of beef steak, for instance, despite causing an extremely high cost to the environment, is enough to provide a good amount of protein.

However, the connection between the number of servings and size may not be clear. The second scatter plot variation removed the size dimension from the first variation. The distribution of the plots reveals only correlations between protein and carbon emissions. The understanding of the number of servings in relation to size, compared to the more straightforward chart, will be tested in a user survey in a future step of the project.

Lastly, the third of the scatter plots series showcased the nutritious qualities of protein and fat grouped along the two axes, represented then in the same manner, while letting the carbon footprint be illustrated through the diameter, a distinct design element. Again, differentiating the categories by color revealed standards, highlighting the high contrast of the environmental cost among them.

Next step
A suggestion received was about adding pictures of the foods to the chart instead of working with sizes. It would be a way to improve the connection between the data and the familiarity of the user with everyday foods. The next step for the series is to select these images, create a visual guideline for them — either in terms of visual aspect or proportion adopted — and add to the visualizations.

Bar charts

Finally, bar charts were adopted in the format of the unit progress bar. In this case, they represent a limited number of portions. The two versions display either the number of typical servings within 100g of weight or 100g of protein. The goal of working with portions was to better connect with how people consume food. These are straightforward charts that could complement the others. For example, to achieve 100g of protein in a day — the average amount recommended for an adult — one would need to ingest 36 portions of pecan nuts!

Next steps
Looking at the chart organized by category, especially the version equalized by 100g of protein, the plant-based group could seem more expressive than the others. However, there are 20 items of it, five more than the other two categories, which counts with 15 each. This unbalancing could be visually misinterpreted. An easy solution to future improvement would be to discard the five plant-based with less protein amount for all the graphs.

Furthermore, the bar charts could also benefit from adopting images, especially if the foods are displayed proportionately. This would enhance the visual comparison of the amounts in each particular typical serving. Also, the number of portions is omitted in the protein variation, and that needs to be fixed.