Gregory Piatetsky, April 21, 2013.

The April 1 KDnuggets post on Data Scientist Hat has generated a lot of comments - here are the selected ones.

It seems that Data Scientists can wear many different hats!

From Tom HC Anderson : a tinfoil hat.

I hope it will help protect data scientists from Evil Radiation, Hype and Overfitting!

From Advanced Business Analytics, Data Mining and Predictive Modeling LinkedIn group:

Vincent Granville: These hats are supposed to protect you or others from something:

• Miner: hat protects head from rock falls, hitting a rock wall
• Cowboy: hat protects head from bullets or from sun
• Baker: hat protects cakes from being contaminated by hair
• Data Scientist: hat protects brain from nefarious radiations and waves

Gregory Piatetsky-Shapiro: Great idea ! We need a data scientist hat that will protect from Hype and Overfitting !

Jilana Rose-Silverberg: Does this mean you suggest a tinfoil hat as the data scientist hallmark?

Gregory Piatetsky-Shapiro: Tinfoil hat, not sure - but a good data scientist hat should protect from Hype and Overfitting :)

Bill Galen: I would suggest a Sam Spade fedora, but that's already been taken.

Panama is a close second--even for the ladies: www.pilotguides.com/images/content/tv_shows/crews/panama_columbia/panama_hat200x250.jpg

Andrew Troemner: This reminds me when I was in grad school for Economics, and one of my old high school teachers swore he would buy me the most atrocious-looking bowtie.

If neckwear is on the table, may I recommend and ascot? Not because it's particularly appropriate but just because I want to be classy while I'm pouring over Python scripts.

Mauricio Pommier Gasser: I like this hat ^ ... sorry... bad joke...

Mike O'Neil: I suggest modelling this with target being value generated over a period with a bunch of hat type variables as inputs. I think there should also be a no-hat variable, and an ordinal scruffiness variable, and a hair-length variable. I doubt the hat-type variables would be significant, but on visual evidence alone no-hat would be significant.

Barbara Lauridsen: How about a data wizard's hat style? A historical version is a tall dunce with stars design. A modern variation could be a floppy hat with huvering holigrams of tiny dynamic dashboard images. What do ya'll think?

Pradyumna Upadrashta: ...they should wear a y-hat

Marshall Presser: A real data scientist wears a helicopter beanie.

Marc de Vries: I would suggest a standard deviation from a normal distributed hat: a SUMbrero

Bill Galen: @ Marc de Vries And there's the Mexican hat filter

Pradyumna Upadrashta: ... it's fairly clear we need to be wearing multiple hats ;-)

Christine Rudolph : What about a hat made by the "Mad Hatter" ?

www.filmofilia.com/johnny-depp-as-mad-hatter-first-look-6456/

Think this decorated with Google glasses, a data glove, a mobile phone, RFID and labs-on-a-chip to be read with small hand-held devices like this:

www.prweb.com/releases/2012/2/prweb9160358.htm

Mad enough ? Definitely fit for the data deluge!