Subscribe
Sign in
pplapi — A virtual database of the entire human population 👥
57
Replies
@rrhoover Super interesting. Did a bit of research -- Apparently it's this PhD candidate's project... http://imiller.utsc.utoronto.ca/
@ggnall I think it's better: these simulated people are plausible, at least in aggregate. It's like lorem ipsum insofar as these data can serve as a "placeholder" but goes farther by also being inherently interesting data (for certain questions), whereas lorem ipsum might as well be gibberish.
@shimmb it is synthetic, which is to say that I am generating it algorithmically, and those algorithms are informed by demographic data (i.e. raw data about the world). There is a plugin system for incorporating various raw data sources and models that feed the algorithms.
@skunkwerk no, you'll need an agent modelling environment for that. There are several examples in the pplapi documentation, but a few names are NetLogo and MASON.
Question: If the data is random how can it be used for prediction or decision making? is it like a demographic "monte carlo" style simulation of population?
@jonslimak The data are not random; they are much more similar to a monte carlo simulation.
@jonathanmarvens it's possible. This is also an ongoing research project for my PhD, so I have no monetization in place. In reality, it would likely be a consulting arrangement rather than pay-for-data, but I'm happy to discuss serious inquiries.
@huangdun I wrote an algorithm for that but it's not publicly visible because it is resource intensive. I found myself in the data at one point last year, but then I made a change to the database, lost myself, and didn't want to search again!
Woah, this is kind of crazy. I'd love to know who made this, the inspiration behind it, and the purpose. Interesting idea that I'd be curious about potential applications of.
They got France localisation completely messed up but apart from that, it's pretty impressive!
I really like this. Especially for randomly populating test databases with quick user info. The 11 - 13 year olds with all this money in their pocket. They are killing the game right now. Kanye always said, "Listen to the kids, Bro!".
@double_are That's part of the idea! This is an active research project, so you've got to be careful with the sorts of questions you ask using pplapi. However, for quickly experimenting, I think pplapi fills an interesting space.
@sridhar_kondoji It's simulated data based on some form of population demographics predictons. They're not real people.
Reminds me a bit of Harin Seldon's psychohistory in Asimov's Foundation series.
@joostschuur @sridhar_kondoji I think it is sortof like a building block of Seldon's psychohistory, yes
Interesting dataset. Cycle'd through 20 or so agents. Many obvious data issues when comparing location, age, and income. (8 yr old in India making 6500 USD; 17 yr old in US making 170k USD, etc.)
First random agent was a teenage girl in India, no internet and has a larger income than me. Great -_-
Product Hunt