Correlation question


Suppose I have:

class % persons % ill persons
a 0% 0%
b 0% 0%
c 20% 20%
d 40% 50%
e 40% 30%

Does it make sense to calculate the correlation between "% persons" and "% ill persons"? My point is that 0% of persons in a certain class implies 0% of ill persons on that same class; therefore, should I exclude such classes (with 0% persons) in the calculation of the correlation?
Or maybe in the calculation of the correlation I could use "% persons" as a weigh. Does this concept exist? A sort of weighted correlation?

Thank you!
Last edited: