[shawn42]

I like your idea here. I think the downside of this approach is you must scale your random data points by a scaling factor to get them in a known range.