As part of FAI-395, data is generated from feature intervals using a normal distribution.

While this is a good first step, fitting a normal distribution might not be very accurate with respect to actual samples. We should investigate ways to generate data in a more faithful way with respect to the existing samples.