I pretty much reject the entire premise of any sort of quantitative "statistical generalization", when it comes to "terror", because terror is an arbitrary qualitative generalization to begin with.
So, here comes this analysis that attempts to state facts as:
1. Terrorists only target civilians
2. Guerrillas only target uniforms holding weapons
...and never the twain shall meet!
These kinds of generalizations in social science are flawed from the outset, but now we try to plot the numbers anyway?
I don't buy into the idea that a heterogeneous mixture of non-state, civilian-killing groups can be boiled down to bald numbers. I think it's a mistake to even try.
I don't think targeting civilians is a rational goal, so how are you going to apply ratios to it?