I figured this would garner enough interest as a separate thread. I have 14,916 profiles from 2014-2018. I'll also poast the profile info I have in my database.

Originally from this thread: http://www.autoadmit.com/thread.php?thread_id=5271892&forum_id=2#45798466

====

During a month-long adderall/vyvanse binge in grad school, I wrote a program that scraped OkCupid profiles in NYC and ran them through a couple machine learning algorithms focusing primarily on pictures. Before OkCupid became tinder-like, I pioneered that shit by training my algorithm on the types of girls I like based on photos alone. Then I trained the algorithm on profiles & questions, and then I messaged the chicks that matched both algorithms. I would hand write those messages and had insanely good luck. But those days are long gone. It's no longer quite that simple, and unfortunately, the best method to get data on someone is to actually interact with them.

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45798489)

Still not the full list, but close enough.

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45798674)

did you fuck any of these girls? i wonder how much dick the average one has run through off the internet.

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45798607)

Do you ever wonder what happened to all these girls?

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45798615)

rahmaxx

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45798739)

rahmaxx: https://imgur.com/a/RXb2SE0

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45798758)

I thought the hot-or-not setup that I used was pretty ingenious: I pretty much copied RankNet and used a binary cross entropy loss to score photos. I'd compare two random photos and pick the one I liked better. I went through thousands of iterations of this and used the person's other profile photos as data points, assuming that if I liked them better in one photo, I'd like them better in all photos against all photos of the "opponent" -- that turned out to be a relatively valid assumption over the years of data I collected.

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45798868)

(we fucked)

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45798770)

Hey ! Im Lauren and i live in the heart of touristy manhattan but

love it! .... My passion is baking cupcakes and i have a lil

business from my apartment... But that doesnt pay the bills so i

have a full time job at a restaurant in hell's kitchen. Im

divorced .. I was married for 3 years and loved it!... But the man

was an alcoholic and it took all my soul to pack up and leave- he

was ruining my life. I kno im young, but i dont regret it one

bit!

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45798789)

I think I may have contacted her once, but she didn't respond. Can't remember.

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45798878)

https://m.youtube.com/watch?v=EZSk_6_V0QA

(http://www.autoadmit.com/thread.php?thread_id=5271987&forum_id=2#45799022)