The Datasaurus Dozen

Pre-order Humble Pi in the USA here:
bit.ly/humblepi
Watch Adam Savage announce his first book club!
• Announcing the Adam Sa...
Huge thanks to Justin Matejka for taking the time to chat with me. They do some amazing work. Follow Justin on twitter!
/ justinmatejka
Read all about Justin's plot and the Datasaurus Dozen here:
www.autodeskre...
Check out the full paper by Justin Matejka and George Fitzmaurice:
Same Stats, Different Graphs: Generating Datasets with Varied Appearance and Identical Statistics through Simulated Annealing
www.autodeskre...
The original Datasaurus plot was thanks to Alberto Cairo.
www.thefunction...
albertocairo.com/
The greater Toronto area had a population of 5,928,040 and mean age of 39.7 in the 2016 census. Check out all the data for yourself.
www12.statcan....
Sorry, Think Maths have not made any teacher resources for this video, but check out their resources for all my other videos here: think-maths.co....
CORRECTIONS
- None yet, let me know if you spot any mistakes!
Thanks again, as always, for Jane Street being my principal sponsor.
www.janestreet...
Thanks to my Patreon supporters who help make these videos possible. Here is a random subset:
Richard Dickins
Kragar
Nikola Studer
Colin Williams
Sarah Gerweck
Ulrich Kempken
Support my channel and I can make more maths videos:
/ standupmaths
Filming by Matt Parker
Editing by Alex Genn-Bash
Music by Howard Carter
Design by Simon Wright
MATT PARKER: Stand-up Mathematician
Website: standupmaths.com/
Maths book: wwwh.umble-pi.com
Nerdy maths toys: mathsgear.co.uk/

[@irakyl]

This video really makes me want to pre-order you new book Humble Pi which is available in the USA on Januray 21st 2020

[@connorking8503]

Joke's on you, I ordered the UK version from MathsGear last year.

[@aperson1]

Wow, interesting comment. I think I'm getting a strange urge to pre-order his new book Humble Pi which is available in the USA on January 21st 2020.

[@ChangingMsagro]

So much so i went back in time and already did it. Maybe i should do it again.

[@gareth2261]

What? I have already read it

[@-perge]

After having witnessed this moving picture, I have recognized within me the pique of interest gained in its topics. Fortunately, my informational thirst predicament is solvable; the advice of our protagonist is here. I shall eagerly, in the immediate future, make it my sole duty to purchase pre-release of the upcoming mathematics and comedy novel "Humble Pi: A Comedy of Maths Errors" written by mathematician and stand-up comedian under the alias Matt Parker. I shall act with haste, since the dawn of release arrives the twenty-first day of the month of January during the twelve-thousand and twentieth year of the Common Human Era, the date and time of said book becoming available for mass release within the territorial control of the United States of America, for if I fail to exchange payment before the aforementioned period of time, I will have failed to do so pre-release. I'm eternally grateful for this video, for its excitability will undoubtedly lead myself toward great enjoyment throught the medium of written text, in both laughter and learned mathematical factoidials. My body drenches itself in sweat by tension created through anticipation toward the arrival of my physical copy of "Humble Pi: A Comedy of Maths Errors" which will be purchasable in participating bookstores across the fifty states and minor outlying territories of the United States of America at a date sometime after the twenty-first day of the first month of the year, that year being the twelve-thousand and twentieth of the Common Human Era.

[@onemadscientist7305]

Or it could just be that there's one person who's 235,343,188 years old and the 5,928,039 others are all newborns.

[@PianothShaveck]

Or maybe there are 10 persons that are 23,534,319 years old and the others are all newborns

[@madjoemak]

Dude you are one mad scientist

[@onemadscientist7305]

@@madjoemak you're goddamn right !

[@John-Smlth]

Or maybe there are 100 persons that are 2,353,432 years old and the others are all newborns.

[@RealCadde]

Or there's one newborn and everyone else is slightly above 40 years old.

[@oafkad]

This actually happened in one of my psychology courses in college. The professor remarked that "The average grade on the last exam was a C." and marked that as a positive because the average person got average marks. The actual numbers were a bit more harrowing. Roughly half of the students had gotten near perfect marks and the other half had flat out failed. You either got it, or you didn't. This was highlighted by college age people literally crying around me after the paperwork was handed out. It was the day I learned about the importance of knowing more than merely the average.

[@oafkad]

For anyone curious, I was in the lower side of things. I actually thought I'd understood the material and even studied pretty religiously. But the exam felt like it covered a bunch of material I'd never seen. Took the course again and did pretty great though. Shame that didn't save my GPA.

[@rolfs2165]

My Basic Electronics prof always provided the graph to his grades. And it always was a bathtub curve. But there wasn't much he could do about it, he already made the material as easy to understand as possible, and used the same problems in the exam as he did in the exercises. It was just one of the courses where you saw whether people picked the right degree or not.

[@riccardoorlando2262]

@@rolfs2165 To be fair, in heavy mathematics-related subjects, either you know the subject and the techniques or you don't. I'm in my last year at uni, studying maths, and as far as I can tell all exam results follow that same distribution. Even my own exam results follow that - either fail horribly, or get excellent marks.

[@HenryLoenwind]

@@rolfs2165 tbh, for a "move on or repeat" type test you want that kind of distribution...

[@aformofmatter8913]

That's why I usually ask for the standard deviation when the prof. mentions exam score averages. Of course SD doesn't tell the whole story, as this video shows, but exam scores are nothing too crazy, so it should give a general idea of how close/far from the average people tended to be.

[@B3Band]

He glossed over this pretty quickly, so I want to restate this: Not only so all the graphs have the same statistical properties, but also _all of the transitions between graphs in the animation as well_

[@olmostgudinaf8100]

That was obvious.

[@LeonardChurch33]

@@olmostgudinaf8100 it may have been obvious to some of us but this might be a helpful clarification for those who didn't immediately catch all the implications of what was being said.

[@olmostgudinaf8100]

Fair enough.

[@guest_informant]

Apart from the occasional annealing?

[@Autom_te]

Yes because it is in fact a visualisation of the underlying (AI) algorithm that reorders the points.

[@magpieMOB]

Only just dawned on me that the acronym for Stand Up Maths is 'SUM'. Better late than never I suppose

[@lyrimetacurl0]

Maybe someone should make a maths channel called PRODUCT as well.

[@JudithOpdebeeck]

Mind. Blown.

[@official-obama]

@@lyrimetacurl0 producing random oblong dark underestimated cold tea?

[@asheep7797]

3 videos earlier, he mentioned it.

[@OberonDam]

1:11 they have the same mean, but they don't mean the same

[@andymcl92]

They have the same standard deviation, but they don't all aww shoot that doesn't work

[@amodgawade4323]

awesome!

[@ThomasBomb45]

@@andymcl92 they have the same standard deviation, but they deviate each other's standard

[@drewdurant3835]

Oscar van Dam you use the best words.

[@TheGreatSteve]

Open the door, get on the floor. Everybody morph the dinosaur.

[@ThomasNimmesgern]

Was (not Was). Yeah. :-)

[@LeonardChurch33]

Justin really likes his keyboards. That is a very custom ergodox on his desk and I see another even crazier one up in the window.

[@blindassassin111]

I thought I was the only one who was going to notice...

[@LeonardChurch33]

I would imagine that in the great Venn diagram of math nerds vs keyboard nerds there is a not insignificant amount of overlap.

[@GregorShapiro]

@@LeonardChurch33 The statistics should be heavily skewed.

[@unvergebeneid]

"Always visualize your data" is certainly good advice if the dimensionality of your data set is low enough but at some point you'll have to use more advanced tools to get to any hidden structure in the data.

[@ninjafruitchilled]

Yeah if your data is only 2D and you don't even look at it then you just need a slap in the face

[@unvergebeneid]

@@ninjafruitchilled true

[@ruroruro]

Exactly. All the Machine Learning and Computer Vision data scientists are quietly crying in the corner, trying to make sense of our 602112-dimensional data.

[@MrCmon113]

@@ninjafruitchilled Bullshit. You assume that everyone working with data is an analyst, who performs data snooping professionally. When you want to solve a general problem, you should actively keep yourself from looking at the data at all.

[@sirbubblegum9224]

The only book I ever read was a short autobiography of a Swedish Comedian and how he met his wife. Never read a single other book ever, not even in school. Making these videos and ending them with "This is in my book" is a really smart move, cause once I know where I can get it in Europe, I'm getting it asap Good job Matt :)

[@DiscountSeanConnery]

The Datasaurus’s favorite method of killing its prey was boring them to death with statistics

[@EdwardNavu]

Added bonus with stealth kills.

[@cachetheline]

So this is actually a great method to morph images. If you generate some class of statistics for one image, then provide an intended distribution of points for the destination image, the mathematical algorithm can adjust the points towards the final image. And of course all the properties of the algorithm and graphical assets can be adjusted and tweaked depending on how you want the process to look, but this actually makes sense for a morphing process!

[@joemyers6016]

Statistics don’t lie but liars do statistics.

[@tjoebtjoeb4080]

Now I understand why my Autodesk license is so expensive

[@Lucas_Simoni]

They are very busy and working very hard at making very important scientific scattersauruses.

[@mannyc6649]

I'm surprised this was presented without mentioning the Anscombe's quartet

[@XaleManix]

I have just started reading your very good book Humble Pi: A Comedy of Maths Errors, in preparation of recieving my pre-ordered copy of your new book Humble Pi: When Math Goes Wrong in the Real World.

[@martinshoosterman]

Ah yeah, I got Humble Pi: A Comedy of Maths Errors when it first came out. Good book.

[@rewrose2838]

They're the same aren't they?

[@rolfs2165]

@@rewrose2838 As much as Harry Potter and the Philosopher's Stone and Harry Potter and the Sorcerer's Stone are the same. (So mostly, but not quite.)

[@cosmicjenny4508]

+XaleManix I bought the adult’s version and the child’s version, just to check that there are no differences, whatsoever.

[@simonjohnston3100]

The star that your square wheeled bike morphed into sure looked like a Parker star....

[@damientonkin]

I did some statistical visualizations for uni last year. Yeah, its a lot harder when you don't actually have the data. They kept doing weird things like subdividing the percentage without explicitly pointing it out so that if you added it up the totals would be 400% or something.

[@sketchyAnalogies]

I’m glad I ordered the UK version awhile ago (though I’m from the states)! Can’t believe you are meeting Adam! How awesome!

[@jakobboers5067]

Victor Mandala I was confused for a while because I have the UK version but it has a different subtitle and cover.

[@sketchyAnalogies]

Jakob Boers that is correct. It looks different, but I believe the content is identical, if not near identical.

[@SaturnCanuck]

Thanks for visiting my home town of Toronto. Hope you liked and please come back in the summer when the weather is warmer. :)

[@iammaxhailme]

Visualizing is important! Once I spent nearly a week trying to figure out why my molecular dynamics code wasn't working becuase it kept randomly either optimizing my molecule into the expected equilibrium geometry or a weird one that I didn't expect until I actually drew out a potential energy surface and realized that the weird one was a local minima, so the code WAS working, I just got really unlucky with RNG seeding

[@EvDelen]

Hey Matt, welcome to Toronto. Hope you have/had a great time here! (Judging by the weather this wasn't filmed on the day it was posted... =)

[@aDifferentJT]

Apart from the correlation you can translate and scale to get any mean and standard deviation you like

[@slendeaway7730]

Super interesting to see the morphs between spread out images and sharper patterns. Seems like points are more likely to 'give up' and get stuck in minima outside the target as well as not confirming to the target perfectly which sorta leaves a shadow of the previous image. I would love to see this but applied to photos just to see the visual affect of changing the pixels but maintaining an average color between two pictures.

[@shermansherbert2570]

Love the vids - new camera looks great, could be just me, but your did your vids sound get quieter? (Maybe too quiet) Would love a gimbal or something for the cam - seemed a little unsteady / wobbly while holding and interviewing.

[@KX36]

I can't agree more with the sentiment of this video. Blood tests are a real world example. One analyser I have spits out about 20 scatter plots and reads 65 parameters off of them, then about 20 parameters and no plots go to the second stage of validation, 12params go out to the doctor and they actually use about 3 of them, making the assumption that all data is perfect. Some of my colleagues advocate skipping the first stage entirely, but it's the most vital!

[@RFC3514]

1:14 - *_"How_* maths goes wrong" or *_"When_* the author gets the name of his own book wrong"...? :-P

[@Ladicuis]

Parker title

[@albertskoope]

"Annealing of data" the corner stone of modern science.

[@zimbabweian]

Can't believe not only were you in my town, but in my very work building!! Sorry to have missed you and hope you enjoyed your stay!

[@CLBellamey]

I could watch those animations for hours...

[@MarcCastellsBallesta]

Yay for the simulated annealing!

[@GoodWoIf]

Hey Matt. Is there any chance you could export your videos with louder sound? On the bell curve of youtubers' video average loudness you're quite far to the quieter end.

[@andriypredmyrskyy7791]

What is up with the lovely hardware on that man's desk? It looks wonderful.

[@green0563]

I know, right?

[@ditzfough]

Split keyboards.

[@hanvyj2]

It's an ergodox. Open source hardware (and software) split keyboard. Made myself one for about £90 or so. I'd recommend it if you do lots of typing though it can be hard to get used to the column layout.

[@Czeckie]

I'm interested in the stuff behind Justin on the shelf. Are those prizes?

[@ditzfough]

Any input on that really long monitor? Make/model? Price?

[@volodyadykun6490]

Parker Square: When Math Goes Wrong in KZbin Video

[@adamsbja]

I am tempted to try and morph the dinosaur into a Parker Square.

[@sas3dx]

@@adamsbja just do it

[@stayawakestudios]

*casually announces crossover of a lifetime with Adam Savage

[@bsharpmajorscale]

Yeah, but when's the crossover with Michael Vsauce?

[@minijimi]

I am going to rewrite in Processing this weekend for my Saturday morning programming project. Seems to be just an interesting adaptation of a genetic algorithm.

[@SimonBuchanNz]

Matt's been getting really mean recently, every video has some standard deviations, but it's been a bit average.

[@recklessroges]

So this is Anscombe's quartet 40 years later?

[@bsharpmajorscale]

This is him now. Feel old yet?

[@gordonrichardson2972]

Reckless Roges A lot more computer power, and animations for those with a short attention span.

[@SimonClarkstone]

To save people looking it up: en.wikipedia.org/wiki/Anscombe%27s_quartet

[@agb2557]

Very cool. And very similar to Anscombe's quartet!

[@ricardo.mazeto]

I've been paying attention to the ratio between dislikes to likes on youtube videos. And good videos have a ratio of about 1/100. This video is one of them.

[@TaswcmT]

I still wonder if the entire difference between the North American version and the original version of "Humble Pi" is that someone spent 2 minutes replacing "Maths" with "Math". Interesting that the title is kept - there is a tendency that stuff have to be renamed to something altogether more violent for the US market: "Math War: Two digits enter, one digit leave! Computing carnage!".

[@benjaminhackett8896]

Humble Pi is out today in the USA! Sweeeet!

[@laurihei]

Just counting the minutes before someone makes the Parker Square dataset.

[@miszamojcyszschmidt1746]

Those are just means, deviations and correlation. First two can be controlled by proper normalisation, and there is left only correlation, that can be easily controlled numerically. I have no idea to why they consider forcing any property hard.

[@mikemhz]

So yeah I guess if you have two points separated whose mean x,y position equals z, then you can move the two points in exactly opposite directions from each other like a mirror image and their mean equidistant position will not change. More data, more options, move two a short distance on one side, move one on the other side a long distance.

[@hotscottrulz]

Man, Justin has a dope computer setup! Those keyboards (including the one in the window) and I need to get me one of those super-wide curved monitors...

[@nymalous3428]

Yay! My copy of your book is coming today! ...I'll have to wait until after work to start reading it... oh well, I guess having a job is necessary... so I can buy your book...

[@macronencer]

Weird coincidence that you released this video just as I was studying Simulated Annealing with a view to applying it to one of my own projects! By the way, please practise holding the camera (phone?) a bit steadier. I felt a little sick...

[@ffggddss]

So having the first 3 moments (N, the number of points, µ, the mean; and σ, the std. dev.) leaves enormous amounts unsaid. But if you include higher moments, it gets harder to draw T. Rex and pentagrams... Still, Matt & Justin are right - there's no substitute for ALWAYS VISUALIZING YOUR DATA FIRST! Fred

[@SamVekemans]

Yeah Toronto! I grew up in the lush vallies of the Sauga. ... Mississauga, which is close enough to Toronto.

[@coryman125]

Reminds me a bit of that principle in cryptography, where you can take the information and process it to get the output, but you can't brute-force process the output to get the starting information again. But as much as I like the idea, I don't know that computer security will ever be a job for Datasauruses any time soon :(

[@andrewwmitchell]

That's quite a LOT of awards on the window shelf.

[@karangupta4978]

To my physics brain, all I see is a bunch of point masses not being acted upon by any external force so their centre of mass remains fixed. So they just move around due to internal forces.

[@williedavis9465]

I bought the book when it was originally released in the UK, it was a good read, would recommend.

[@andrewseburn]

Woot! Another video in quick succession! Thanks for the hard work to make it happen Matt! PS. I'm excited that I pre-ordered your book! But I'm disappointed were in my neck of the woods without me knowing ;) haha Maybe someday....

[@Wyvernnnn]

What if you wanted to switch from shape A: '| ' (one | on the left) to shape B:' | |' (two | on the right), wouldn't all the points get stuck to the first leftmost 'pipe' (|) during the transition from A to B and not reach B's rightmost pipe?

[@sayanghosh6996]

Now this channel is just a promotional page for your book

[@JosuaKrause]

Francis Anscombe came up with that in 1973: en.wikipedia.org/wiki/Anscombe%27s_quartet

[@deanc9195]

Autodesk is the best 2-faced company... Creating envelope pushing engineering stuff & paying folks to make graphs that look like dinosaurs. GOOD JOB, AUTODESK. I’M SORRY I CANT AFFORD YOUR SOFTWARE.

[@peterlewis1554]

Do higher dimensions make it easier or more difficult to perturb data while preserving various statistics? I would guess the former since you'd have more directions to wiggle, but I've been fooled by intuition before.

[@bobshewberg]

Is he using Markov-Chain Monte Carlo to get new shapes? That's awesome! I'll definitely be messing with that code

[@awesomedavid2012]

The color grading in the beginning has be astonished. I mean I'm no professional but in my amerature opinion, it's very nice 👌

[@beefytaquitos]

If there's any way you can bump up the volume of this video, it's ridiculously quiet and the only way to hear it with anything else going on the PC is to turn everything else down or mute it entirely.

[@sam08g16]

Posted 49 seconds ago, ok I'm officially spending way too much time on KZbin.

[@TheMrvidfreak]

On the contrary, 49 seconds too little!

[@Thelondoman]

Matt How to fold perfectly any paper in three equals parts avoiding the "accordion" method that is not too exact, and get a mark or some origami /math technic to get a Third of a distance only folding the paper

[@HagenvonEitzen]

Incidentally, this is solved in my highest voted answer on math.stackexchange: math.stackexchange.com/questions/736346/how-can-a-piece-of-a4-paper-be-folded-in-exactly-three-equal-parts/736351#736351

[@JNCressey]

If you don't mind a ton of extra folds: Make an arbitrary line. Make an arbitrary segment on the line. Use the segment to make 4 equally sequenced points A B C D. Make a line from one corner through A, and another from the adjacent corner through D. From where they cross, project B and C to the edge. Repeat for oposite side. Use these points to position your folds.

[@JeroenBouwens]

There have been several times in the past when proposed a legal ban on supporting an argument with some average number without also mentioning the corresponding standard deviation. Nobody was ever interested

[@ruben307]

well it seems they had no reason to be interested

[@vleessjuu]

Standard deviations can be just as deceptive as means. After all, it's again just one number than can be generated in many different ways. And for data from some distributions, means and standard deviations aren't even well-defined do not summarize the distribution in any helpful way. Turns out that dealing with data, probabilities and uncertainty is hard.

[@Wordsnwood]

Yeah, sorry, not preordering --- already bought it from Maths gear last year, on PI day, am I forgiven? BUT THANK-YOU also for coming to Generator 2020 in Toronto. I was SO pleased to see you there.

[@danielrhouck]

I wonder how much more difficult this would be if you added more stats: higher-order moments; median, quartiles, percentiles, etc.; and other things that measure the relation between X and Y besides correlation (don’t know what the standards are but I’m sure there are some).

[@j.kaimori3848]

I think that's the point, but even with more measurements you could still get two different styles graphs similar statistics.

[@mrguffaw]

How did you get through this entire video without mentioning Anscombe’s Quartet?

[@MrBleulauneable]

And no mention of it in the top comments yet -> up you go !

[@SimonClarkstone]

To save people looking it up: en.wikipedia.org/wiki/Anscombe%27s_quartet

[@BryanLeeWilliams]

Glad I have my notifications on. I never see your videos in my subscriptions.

[@matthewellisor5835]

I think that depiction of a bicycle has a case of Parker-Wheels. Now, thinking of it, the images reminds me of a dream in which I have this strange feeling that someone in my household will pre-order a certain book (available 21JAN2020) with "Humble Pi: When Math Goes Wrong in the Real World" as the title.

[@bordershader]

2:00 *massive monitor envy* man I'm so in awe!

[@witerabid]

There I was using AutoCAD for a year and a half and I never knew Autodesk is a Canadian company... You learn something new every day. 😅

[@bazahaza]

It's not. They are headquartered in California. Its U.S. locations are California, Oregon, Colorado, Texas, Michigan, New Hampshire and Massachusetts. Its Canada offices are located in Ontario, Quebec, and Alberta. Even more worldwide.

[@witerabid]

@@bazahaza Ah, ok. Thanks. 🙃

[@atimholt]

My dad pre-ordered Humble Pi for me for my birthday. Read it in two days. I found it very instructive! I love the embossed pi under the dust jacket, but I’m not sure whether I can trust it. (Love that guy’s keyboards.)

[@normalemoji]

We don't agitate the dots. Who's agitating my dots?!

[@MonsieurBiga]

The 21st of January is my birthday! 🎂 Love you

[@seanli7493]

Those custom mechanical keyboards are making me jealous.

[@petemurphy7164]

The book must be released a bit late in America. I read this months ago (and would Strongly recommend it)

[@MCLegoboy]

But what about transforming the dinosaur into other dinosaurs?

[@Frrk]

Autodesk? Never heard of it! Must be a small, insignificant startup company /s

[@omikronweapon]

Matt Parker combined with Adam Savage? I'm not sure the fabric of the universe can withstand that amount of sheer optimism.

[@matthewgiallourakis7645]

I finally bought Humble Pi on Audible! Awesome!

[@charleswang833]

Couldn't you just linearly interpolate the data points and then use a translation and a linear transformation to keep the mean and covariance data the same?

[@LounoirRecords]

ah the fun times of averages, i find it amusing that half of toronto are all 80 year old couples who all had babies at age 80 that's just brilliant

[@microtubules]

Next time you visit MaRS, come up to the 15th floor! We don't do maths, but we can show you how your cells behave.

Simulated annealing! I miss my uni days, I was very much interested in those methods back then.

[@gib20]

Cool keyboard. I think it's an ergodox?

[@aaronsmicrobes8992]

Can confirm, that's an ergodox. I own 2 of them.

[@jimi02468]

@@aaronsmicrobes8992 That makes 4 keyboards overall

[@drewdurant3835]

You make math fun!! I wish I had teachers like you.

[@clusterfork]

I wonder if it is just as easy to do this when keeping the entire covariance matrix constant (not just the standard deviations in the individual directions)

[@vleessjuu]

I believe that the correlation coefficient is kept constant between the pictures as well; they just didn't bother to state that in the video. At any rate, as long as the number of degrees of freedom is much greater than the number of statistics you're keeping fixed it's pretty easy to do things like this even with higher-order statistics.

[@HackSawSees]

Need one in the shape of a Gaussian distribution.

[@chainmaillekid]

What are the differences between the American and original releases? Fewer S's?

[@georgettebeulah4427]

Pie charts make so much sense I can relate to it but do not understand it or any of it at all so I can get your meaning but will like to learn more.

[@nickcook2775]

Where did you get the CAD files for the solids and shapes of constant width from? I've been wanting to print them but have had little to no success insofar.

[@ericveatch9819]

Just got my copy of Humble Pi from Amazon. Alright so the page numbers are in reverse order, perfectly normal. I open to the first page. Left side page, completely blank. Page number: 314. Of course.

[@XxRiseagainstfanxX]

My first propability lecture ever had a theorem called "Noninformativity of the mean" lol.

[@MrLrodlima]

Fantastic work

[@masheroz]

So, anscombe's quartet?

[@shambosaha9727]

Multiply that craziness by 3

[@masheroz]

@@shambosaha9727 but how do you do a video on datasets with identical parameters without at least mentioning anscombe's quartet? The transitions are cool, the myriad shapes are cool, but it builds on the original idea from anscombe.

[@shambosaha9727]

@@masheroz Yep

[@bocaccio5]

While these data displays are cool (and I agree that people should check their scatterplots), the presentation of results seems a bit disingenuous, particularly with respect to the correlation coefficient. The means and SDs correspond to each of the marginal distributions, so it is very simple rearrange the points in the two-dimensional space without affecting these statistics. Maintaining the correlation under these shapes is a bit more challenging; however, depending on the actual statistic computed (Pearson, Spearman, etc.), the bigger question is whether the underlying assumptions are satisfied to support the appropriateness of computing the given correlation?

[@peema10]

No HB Plane morph? I'm vaguely disappointed.

[@leikamcraft7275]

Pre-ordered the book! I'm excited :)