
Ed Yong asked his legion of Twitter followers to explain Principal Component Analysis (PCA) in a single tweet that would be understandable to a schoolchild, here is my Storify of the results.

RT @edyong209: Explain principal component analysis to a schoolchild in a tweet. Go.

@edyong209 All you want for your birthday is… A quick n easy definition of principal component analysis! 🙂

His followers responded, and the results were interesting, initially they ranged from vague to overly technical

@edyong209 PCA – Basically factor analysis yeah? Principal component analysis?

@edyong209 “If you have to talk about something really complicated but you can only use two or three numbers, what do you do?”

@edyong209 Taking lots of numbers and boiling them down into as few numbers as possible.

@edyong209 A colleague once summed up PCA as a way of letting data decide for itself which patterns best describe it.

@edyong209 PCA is a way to explain the relationship of things that at first glance appear completely unrelated from each other.

@edyong209 My short answer: it’s used to ID correlated variables that are being influenced by the same underlying phenomenon.

@edyong209 PCA is a German diagonal iteration reducing data noise to a murmur

@edyong209 If data shows a trend, PCA can work out which of a list of factors explain most of the trend. [ >30 characters left! 😉 ]

Those tweets tended to forget about “explain to a schoolchild” aspect of the challenge, others picked up the slack

@edyong209 compare the height, eye color, hair length, etc. of the people around you. Where do you notice the most difference?

@edyong209 It’s like representing a road with a line on a map. The road has width and length, but the length is sufficient for most purposes

@edyong209 Every politician has lots of beliefs, but you can guess most if you know whether politician is liberal or conservative (1st PC)

@edyong209 PCA tells you which features best describe the differences between items in a group, like colour in a box of lollies

@edyong209 Imagine collecting all the rocks in the yard, but you can only keep 10. PCA picks the ones with the most common color & sizes

@edyong209 Best soccer teams: do they score more or have the best defense ? PCA ranks values of teams based on multiple data of this kind.

RT @holy_kau: @edyong209 PCA is like picking cookies with the most chocolate chips and seeing where they came from in the jar.

RT @APV2600: @edyong209 Imagine lining the crayons in a box up based on their color similarity. Now, in the other direction, separate them by the wear.

I personally like this video explanation, which demonstrates that PCA can be a fairly intuitive concept

0
likes·
0
comments 
Some of the more bizarre answers tended to relay the mystery and seemingly magical side of PCA

@edyong209 Principal component analysis, kids, is much like a stranger bearing sweets – to be avoided at all costs. Don’t believe me? Try it

@edyong209 Measure everything you can. PCA will turn it into fuzzy balls.

@edyong209 Something nerds do to show off and that the rest of us secretly don’t understand but always look knowingly when it’s mentioned.

@edyong209 PCA compares apples with oranges in a fruit salad kind of way.

@edyong209 PCA is what you do when you have lots of numbers, but dont know what to do with them 🙂

@edyong209 PCA is black magic math that tells which of dudes are nerd, fun, prep or jock & explains what makes them nerd, fun, prep or jock

@edyong209 it’s all mathemagics to me. #NeedPCAforDummies

@edyong209 Like when parents try to describe something they don’t completely understand, like sex, and draw vague diagrams that explain 27%

As the challenged progressed, Ed Yong retweeted some of the best responses, with equal measure of the good and the bizarre

RT @edyong209: Some of you are kicking ass at the ExplainPCAtoaschoolchild challenge. Others have bizarre ideas about what schoolchildren are like

RT @aimeemax: @edyong209 PCA is a way of finding basic similarities between many different things – linking complexities by their simplest similarities

RT @mickresearch: @edyong209 PCA of leaf shape would first give size, then fatness (width:length), wiggles of leaf edge, etc. 1st parts give largest variation

RT @CorrinneBurns: @edyong209 PCA uncovers hidden relationships between things. For example: everyone who follows @edyong209 hates cheese, and is insomniac.

RT @VinJLynch: @edyong209 PCA asks data how it wants us to describe it to our other friends.

Some noted that his retweeting behavior, was itself an example of PCA

@edyong209 PCA is like RTing only the best explanations on Twitter.

@edyong209 Great, you just did PCA–you summarized many diverse tweets with two axes of variation (PCs): asskickyness and bizarreness.

The end result of the challenge was a little bit of education and interest in PCA, with some mathy/nerdy entertainment thrown in for good measure

I’ve enjoyed seeing responses to @edyong209‘s PCA definition question get weirder/more entertaining over some hours! twitter.com/VinJLynch/statu…

.@edyong209 Your request to explain PCA to a child reminds me of the Flame Challenge: flamechallenge.org

@edyong209 I am a teacher and I have LOVED reading how your Followers responded to that PCA problem you set #inspirational

After all those @edyong209 retweets, I really want to learn more about PCA
Pingback: Eigenwhowhats? The great Twitter PCA challenge!  Appetite for Awesome
Pingback: I’ve got your missing links right here (22 December 2012) – Phenomena: Not Exactly Rocket Science
Pingback: Stuff we linked to on Twitter last week  Highly Allochthonous
Pingback: RoundUp Ready: A Year in Blogging Edition  On a QuasiRelated Note