Recently a zoology study about dog defecation has gone viral. No, seriously.
In an attempt to test dogs’ sensitivity to Earth’s magnetic field, researchers recorded the urination and defecation of a sample of dogs and plotted the angular position of their bodies during these acts against the magnetic field. The results of the study produced great visualizations like these:
Each dot represents the mean angle of a dog across all of it’s observations. I believe the dotted line through the middle represents either the mean of all observations and the arcs some sort of confidence interval or magnetic field spread, but I am not certain.
I reproduced the chart in QlikView using radar charts, with the dotted line as the mean and the highlighted arcs as a 95% confidence interval. Random data was used to produce the following:
The creation of this chart required several elements:
- loading the inverse of each point; for example, if an event happened at 0 degrees, then a duplicate needed to be created at 360 degrees
- grouping the degrees into bigger buckets in order to plot dots. In this example, I grouped them by units of 5
- creating text labels for only 4 points
- this was accomplished using the a dual function to apply a blank label for every point except the 4 I wanted to show:
dual(if(mixmatch(Degrees,0,90,180,270),Degrees & ‘°’,”),Degrees) as DegreesLabel
- creating a second radar chart in the background for plotting the mean, confidence interval, and complete circle
- the background chart required an isolated set of data that covered all degrees. I autogenerated numbers from 0-360 into an isolated table and then used expressions and variables to add my formatting
- Mean dotted line expression
=if(AllDegrees=mod(round(vMean),360) or AllDegrees=mod(round(vMean)+180,360),1)
- Outer circle expression
- Color expression
=if((AllDegrees>=round(vConfidenceIntervalLow) $(=if(round(vConfidenceIntervalLow)>round(vConfidenceIntervalHigh),’or’,’and’)) AllDegrees<=round(vConfidenceIntervalHigh)) or (AllDegrees>=mod(round(vConfidenceIntervalLow)+180,360) $(=if(mod(round(vConfidenceIntervalLow)+180,360)>mod(round(vConfidenceIntervalHigh)+180,360),’or’,’and’)) AllDegrees<=mod(round(vConfidenceIntervalHigh)+180,360)),rgb(100,100,100),rgb(200,200,200))
- this expression looks pretty complicated but it is just peforming a simple check to make sure it calculates the ranges correctly. The problem with the range calculations is when they overlap 0 degrees, since the numbers reset. For my implementation I used a logical condition to switch out an “and” or “or” statement, but there may be a better way to solve this problem.
Instead of posting a link to the QVW here, I’ve decided to give Stefan Walther’s idea a try and post the document to GitHub. The script is kept externally from the QVW so that changes can be tracked separately from the front end.
Get it here.