Let's say I have a set of data which consist of Real Value, Approximation I, Approximation II. For instance, the data looks like this:
Now, I'm not sure which error to choose so one can infer that Approx. II is better than Approx. I. What I did so far is calculate the relative error for each points (which is a lot), then say that
$\dots$ by observation, we could see that approx. II gives lower relative error than approx. I $\dots$
The thing is, while the majority is true, at some points, approx. II gives higher relative error. So I couldn't say that the statement is entirely true.
Are there any error, as a single value, that could represent this kind of data? I've been looking at SSE (Sum Squared Error), but I'm not sure it's what I need...
Any insight would really help, thanks beforehand.