The Big Data of Big Hair (2019) | Hacker News Reader

The Big Data of Big Hair (2019)

pudding.cool

86 pointsEndXA12 comments

I still love both the look of big hair and they smell of AquaNet. Turns out the data proves it: right when I first starting noticing girls those two things were at a maximum!

That’s so cool! It’s amazing how much we can do today with Big Data. But I personally believe that the best use of Big Data is in the finance sector. Why? Cuz there is so much data to track and analyse and according to this article the possibilities are limitless: https://devsdata.com/big-data-financial-services/

I showed my wife this article and she immediately pointed out the prevalence of "bonus hair", i.e. extensions of one sort or another, which was not as available decades ago.

Men's hair increased until 1979, when the movie Hair came out, then started dropping.

Seems the movie coincided with (or caused?) Peak Hair.

I want it long, straight, curly, fuzzy, snaggy, shaggy

Ratty, matty, oily, greasy, fleecy, shining, gleaming

Streaming, flaxen, waxen, knotted, polka dotted

Twisted, beaded, braided, powdered, flowered and confettied

Bangled, tangled, spangled and spaghettied

Very fun application of big data and I love the choice of the source itself. I was puzzled at first how one could accurately assess the popularity of a hairstyle because, depending on the type of person asked, the result would be skewed. But this is smart because everybody got their best look for the yearbook photo (however regrettable that look may be in the future is debatable) so anybody who liked big hair at that moment would have it for the photo. Plus, yearbooks were collected from all over America so there's less chance of the data being skewed toward the 'hipper' states. (I'm not an expert on which states those would be but I'm presuming places like California and New York are closer to the top in that ranking?)

> But this is smart because everybody got their best look for the yearbook photo

Ugh. Thinking back I remember quite a few years where mine was particularly bad because my folks always wanted me to get a haircut right before yearbook photos...and a haircut usually meant my dad exercising his questionable scissor-and-clipper skills in the basement.

Bad bowl-looking cuts when I was really young (just comb it down and I'll cut whatever hangs past your eyebrows!) and stair-step "fades" on the sides when I was a little older (never learned how to blend so it was just switching up the guard as you go up from ears to the longer parts on top.)

I can laugh about it now because I also remember "graduating" to cutting my own since at least then, the questionable results were my own fault. It's definitely not easy even with multiple mirror angles and a bit of practice. I did get decent at it by the time I was old enough to just pay the $20 every month for something that actually looked good.

Now my partner is a stylist and she gives me free haircuts in the kitchen whenever it gets to the point where I start bugging her again. I'm pretty sure my dad doesn't take offense that I've switched barbers.

Very cool. Fun use of models. Question: what did you use to generate that SVG chart?

Thanks! charts were built in D3 and then rendered out to video using this tool: https://github.com/russellgoldenberg/render-d3-video

I'm really impressed at the level of attention paid to the data at each stage, to ensure that junk values weren't being propagated in the analysis. I wish ML papers aspired to this level of quality!

Very cool! Nice dataset selection. Could you possibly use this as a classifier to determine the year of a given photo?

One small quibble: this is more of a Computer Vision work, not Big Data. Great writeup nonetheless

A relevant short video from the article is available here: https://www.youtube.com/watch?v=fzwrpAfMdKw

OK but what's the unit used in the chart?

Reply on news.ycombinator.com