-By using Euclidean distance as the metric we observed the dimenion to be 2 if we take the neighbourhood of the pixels(K) as 5. This is because there is a steep change in the reidual variance of the graph as we go from 1 to 2 but there isn't much change in going from 2 to 3.
-Using low cost euclidean function is a nice approximation as the change in the arm position is very less.

-When we take the value of K = 7, the irregularity is because the rise in residual variance after dimension 2 as shown in the results.