News for hackers. Stuff that matters.
Re:Unisex Names – Data Analysis Use Case by jackcarter on Saturday June 20, 2015 @03:28PM (#9750801)
>Let’s say we define 20%-80% as our threshold, so if a name is used on, let’s say girls, more than 80%, we call the name female-dominant. If it is used on less than 20%, then we call that name “male-dominant” name. Otherwise, we are left with the range from 20% to 80% (60% of all names) which I propose to consider “unisex” names.

Interesting writeup, but this is obviously faulty reasoning.

Re:Unisex Names – Data Analysis Use Case by lmitchell on Saturday June 20, 2015 @03:44PM (#9750840)
It's pretty arbitrary, yeah, but why 'obviously faulty'? Seems to me like it's just a different threshold than you or I might have picked, I can't think of a better metric to judge by though.

It would be interesting to see which names were the 'most' unisex, though - ie. those with closest to 50% spread, regardless of popularity.

Re:Unisex Names – Data Analysis Use Case by kevb on Saturday June 20, 2015 @04:05PM (#9750894)
The "60% of all names" is wrong. Let's imagine there are 3 total names, Alice, Bob and Casey. Alice is > 80% female, Bob is < 20% female and Casey is 50% female. Even though we're using the 80/20 cutoff, only 33% of the names are unisex, not 60%. You can't just subtract 20 from 80 to get the number of unisex names as the distribution is unknown (or unmentioned). You'd have to actually measure or take the distribution into account.
Re:Unisex Names – Data Analysis Use Case by pedrosorio on Saturday June 20, 2015 @05:13PM (#9751065)
Yes. I find it hard to keep reading after that statement.