# Using Axis Equal for Matlab simulation plots of the SIR model gives very flat solution curves

My professor taught us to always use the Matlab command axis equal for our simulation data plots - for all of our homework assignments.

However, in studying the SIR infectious disease model, I'm studying a scaled version, so instead of using population counts I am more conveniently using fractions of populations, so that my vertical y-axis ranges simply from 0 to 1, while the x-axis is for the time domain. Using this approach, and the command axis equal, my data plots look like almost flat lines. When I remove the axis equal command, the solution curves look much closer to the ones given in the news reports, YouTube videos, etc.

My question is: Should I always be using axis equal? And if so, how do I fix my plots so that they look like what we expect to see?

• Why do you think that it's important to use equal axes for quantities whose dimensions (fraction infected vs. time) aren't comparable? Apr 17 '20 at 5:41

In general, one should be very cautious in taking any blanket-statements as an absolute rule. While your professor before might have implied a different logic sequence, mentioned/forgot to mention some details, was wrong or did have only a certain use case in mind – are among possible explanations.

axis equal can be good for some cases, and might be a go-to default for certain kinds of assignments. However, tuning and customizing visualization are also an integral part of scientific research, and proper scales, ticks, grids, view angles, labels are part of that. These tools can be used for both a good purpose (better visualization of a certain phenomenon, a clear indication of a trait) and a bad purpose (deliberately confuse the audience, hide certain aspects). It is important to stay on the "good side".

To sum it up: no, you certainly don't have to use axis equal for all visualization, especially when different axes have incomparable quantities, or even if the scale of the quantities that are comparable are very different.

I would also suggest reading the research papers in reputable journals from your research area. Also, if you are interested in data visualization, I highly recommend reading E. Tufte "The Visual Display of Quantitative Information".

• @user35665 you should trust your judgmenet as an honest researcher. If Matlab's auto-scale helps – go with it. Sometimes it is, sometimes it is not. Apr 17 '20 at 20:09
• @user35665 ALWAYS - is too strong word. In some cases, scaling from 0 to 1 can be beneficial. But I would never agree it is the case for all things. Apr 17 '20 at 20:10