Data Analysis by Driven IFS

Sample

For differences of data, another natural division scheme is to take B₂ = 0 and B₁ and B₃ some fraction q of A = max{|x₁|, |x₂|, ...}, the maximum distance of any data point from 0. That is,

B3 = q*A

B2 = 0

B1 = -q*A

Unlike equal-size and equal-weight bins, there is a whole family of zero-centered bins, depending on the value taken for q. Different choices of q emphasize different aspects of the data.

Here are driven IFS with three choices of q. Click the picture to enlarge.

q = 0.5	q = 0.25	q = 0.1

A more subtle problem is to quantify how these driven IFS change as q ranges from 0 to 1.

Return to Sample.