I used this at one point when I needed a large synthetic "customer" dataset, and found the sample size to be somewhat limited. I had to munge first/last names w

I used this at one point when I needed a large synthetic "customer" dataset, and found the sample size to be somewhat limited. I had to munge first/last names with added random characters. But looking at it now, it appears that's improved. For example, there are about 3000x500 first/last name combinations, for a total of about 1.5M possible full names (for en_US). However, if you need, say, as many unique last names as possible, 500 might still be too few. Still, a nice project.