Hi, I am trying to correlate two curves with same x values and different y values? The curves shouldnot be correlated based on the data point distance but base on its trend or in short its shape.
is it right that it should be unimportant if there is a great distance between the values on the two curves, but it is important if their shapes are similar?
So, what's wrong with a simple regression? It's coefficient should be high poisitive, when high values on the one curve get along with high values on the other curve and when low values on the one get along with low values on the other curve. And this should indicate similar shapes, should'nt it?
Another idea is to compute first difference scores and to compare them between the curves. The differences should reproduce the form of the curve.
I'd suggest using Pearson coefficient (https://en.wikipedia.org/wiki/Pearson_correlation_coefficient). For example we have two lines below, the resulting Pearson coefficient for them is 0.763 which means that data is correlated or lines are close to each other.