What is the formula that Phoenix uses to identify outliers in a box plot?
The formulas I use are as follows:
Below Q1 - (1.5 * (Q3 - Q1))
or
Above Q3 + (1.5 * (Q3 - Q1))
However, when I apply this calculation to my data, I get different cut off values than what Phoenix appears to use.
Is it possible to customize the outlier cutoff values used in the box plot?
Did you see the annotated diagram on page 86 of Phoenix 1.4 Data Tools and Plots Guide.pdf ?
You can change the degree of ‘outlierness’ by changing the IQR from the default (1.5) to e.g. 3 to show ‘severe’ outliters.
If that doesn’t clear up your problem, could you perhaps post your project and an example of where you calculations differ.
thanks, Simon
No, I hadn’t seen that reference. Thanks for pointing it out to me. Quite helpful.
After glaring at my data for a while and trying to understand why the math wasn’t working out given that Phoenix and I appear to be using the same equation, I finally realized that two outlier values in the data set are very close together and so they appear as a single point on the graph. In the box plot, it appears there are only three outliers though in the dataset there are clearly 4.
I am attaching the data in case anyone else wants to plot it and be mystified
Test Data.xlsx (7.63 KB)
Oh yes I see 3460 & 3450, tricky to know how to handle this… their ‘marker’ does appear more ‘bold’ than the other two. An offset/jitter is how I might try to handle this but we don’t support it in the Phoenix box plots currently.
Is it real data ? or a test set?
Simon
