Averaging component-wise would certainly be one meaningful way to do it, as you have done. If you want to weight one point more than others in your averaging, you can also do that. I'm not sure if I know of other, meaningful ways to do this.
Hello guys,
I have a set of n, k-dimensional vectors. I want to find (approximate) the center of this data (not necessarily a vector that belongs to the set).
Lets say the vectors are three dimensional points. Lets say I have five of them as such:
p1[1,-2,3]
p2[2,5,-3]
p3[3,3,-1]
p4[0,5,2]
p5[1,-7,-4]
c = (p1+p2+p3+p4+p5) / 5
c = [(1+2+3+0+1), (-2+5+3+5-7), (3-3-1+2-4)] / 5
c = [7, 4, -5] / 5
c = [7/5, 4/5, -1]
Is this a correct approach? Does it make sense?
Any other (faster) ways to do this?
Thank you!
Averaging component-wise would certainly be one meaningful way to do it, as you have done. If you want to weight one point more than others in your averaging, you can also do that. I'm not sure if I know of other, meaningful ways to do this.