Wikipedia:Reference desk/Archives/Miscellaneous/2019 July 23

= July 23 =

What's the closest I can get to something that spits out the home addresses of n random Americans?
For whenever I wonder what the graph of distance to the nearest food/roof height/plant hardiness zone/etc. by percentile looks like. No names, numbers or anything, just 100 buildings where a building with 100 Americans is 100 times more likely to be picked than a building with 1. If this is unavailable, are there proxies that don't grow the error bars much?

Also how do I draw error bars for the n samples? I dropped out before being taught this in school so I have no idea if say the lowest sample should be given a 1- or 2- sigma error bar that goes up to the value of the 20th sample or to between the 5th and 6th sample or what. Sagittarian Milky Way (talk) 22:53, 23 July 2019 (UTC)


 * The generate.plus website has a real random address generator. It gives you a real address, but randomly chosen. If an address has multiple units (such as an apartment building), it is more likely to appear because each unit is an address. The problem is that the tool generates one address per click of the "generate" button. 135.84.167.41 (talk) 13:42, 24 July 2019 (UTC)


 * Thanks, that seems good enough. I suppose I could count a sample who's census block has say 1.1 or 0.8 times the average humans per household as 1.1 or 0.8 samples but that sounds painstaking. Sagittarian Milky Way (talk) 00:14, 25 July 2019 (UTC)