This week, I took Dr. John Rasp’s dataset on bodyfat percentage sampled from 252 men, along with additional measurements of body size including chest, abdomen, arms, legs, neck, etc. The data states that body fat percentage is “normally measured by weighing the person underwater – a cumbersome procedure.” The data aimed towards estimating body fat percentage with these alternative body measurements which are much easier to collect. The data were taken from the Journal of Statistics Education website.
I utilized the free online data visualization tool Plot.ly to create a scatter plot showing the relation between body fat percentage and body weight.
This visualization allows the viewer to not only see the data that was collected, but also see a relationship between the variables selected–something the table alone does not show. The x-axis shows the body weight measurements of the sample of 252 men, and the y-axis shows their calculated body fat percentage based off measurements collected and calculated by the researchers. I made this into a scatter plot, as any other plot did not seem effective in accurately representing the collected data. By adding a best-fit line, the plot indicates a positive relationship between body fat percentage and body weight. I also went ahead and made another scatter plot to visualize the relationship between the estimated body fat percentage and thigh circumference (in centimeters), one of the many various body measurements.
Again, the plot shows a positive relationship between the variables.
All in all, this data visualization allows users to clearly see positive or negative relationships between selected variables, or if a relationship is present at all. I learned from this exercise is that one should be mindful of which variables to represent on each axis and to careful evaluate whether or not one’s selections are relevant to what one is trying to observe. I ran into this problem a few times, and ended up with plots presenting contradicting relationships.