KCOCCO ~ photo

Coding Avalanche

Kevin Cocco — Thu, 02 Feb 2012 08:26:05 GMT

Awesome ECT Results from trentmeisenheimer on Vimeo.

Scary backcountry conditions this year. Nice video Craig, love the slow'mo'audio!
Good weekend to write code. I am pitching PredicTag at Salt Lake City StartupWeekend

IBM's Watson - Machine Learning

Kevin Cocco — Wed, 01 Feb 2012 19:05:39 GMT

Cool graphical video of how Watson answers Jeopardy questions. Here is also a TED talk about Watson's win.

Google Refine and Translate

Kevin Cocco — Sun, 29 Jan 2012 23:20:54 GMT

Wondering what Japan is saying about Fusion-io?
Tweets can be hard enough to decipher let alone translated tweets...

Here are some tips on translating Tweets with Refine and Google Translate REST api.

- Imported data into Google-Refine remember to specify utf-8 character encoding on import.
- You need to get a Google Translate API key
- Tweet text column drop down select Edit column - Add column by fetching URLs...
"https://www.googleapis.com/language/translate/v2?key=&target=en&q=" + escape(value.substring(0,140),"url")

- The above creates a new column with the Translation json, now to parse, Use the add column based on this column command.
- below gets the text
value.parseJson()["data"]["translations"][0]["translatedText"]
- below get the predicted Language code
value.parseJson()["data"]["translations"][0]["detectedSourceLanguage"]

Some very wildly different results when comparing Twitter User Language, DataSift Prediction, Google Translation Prediction... check out the Google-refine text facets below:

Got Dirty Data? Google Refine it!

Kevin Cocco — Thu, 26 Jan 2012 22:32:21 GMT

If you work with data, json, XML, especially dirty data you must check out
Google Refine. Thanks Google!

Watch part 2 & 3 for the real fun....

Fusion-io Brand Tracking

Kevin Cocco — Wed, 25 Jan 2012 23:13:39 GMT

Tough day of trading for FIO 25.50 -4.84 (-15.95%)!
I started my brand tracking data capture for FIO today, capturing all(yep all) Tweets, FB posts, a few others... Interesting day to start following Fusion-io Q2 results last night.

Tweet from highest Klout score of 61 with 9,409 Twitter followers:
regvulture - The Register:
"Fusion-io revenues flash upwards: Market punishes Fusion for margin fall. Server flash array vendor Fusion-io saw http://t.co/YNbdov3G"

Tweet from user with most Twitter followers: 11,331 with a Klout score of 52
SeekingAlpha
"Disappointing Margins Crush Fusion-io: Storage Wars Heat Up http://t.co/EZyZnyBR $OCZ $STEC $FIO"

Each article mentions some possible FIO competition. The Register mentions "EMC's Project Lightning" and SeekingAlpha "OCZ Tech (OCZ)". I wonder who may be caught in Innovators Dilemma? Is EMC really not putting their A+ team and priority into Flash because their MBA's made a graph that showed this cannibalizing their core business model? If EMC doesn't cannibalize/disrupt their business model someone else will. Is Fusion-io brushing off lower sticker cost OCZ Flash solutions because they don't have the fastest X per Y performance and only meet a niche that is not as profitable...now? OCZ < Fusion-IO < EMC =?

I am personally still bullish* on FIO. The public company curse of balancing cashflow for public display every 3 months must be a nightmare for a startup. Brand building, engaging new clients and verticals in proof of concepts, tweaking manufacturing for new product... I would expect these investments to pay off.

Data capture running, predictive model building starting soon ...

* I am CS guy not stock guy, I own shared of FIO

Chris Sharma, Kevin Cocco & Jeremy Jones

Kevin Cocco — Mon, 23 Jan 2012 01:59:51 GMT

Had the amazing luck to run into two of my favorite athletes of all time at the same time during the Salt Lake City Outdoor Retailer show. Arguably Chris the best climber and Jeremy snowboarder of all time.

Jeremy Jones

TGR - Deeper Trailer from IronFistMedia on Vimeo.

Chris Sharma
Sharma Video

LucidCharts + Wordle = Google Prediction API Graphic

Kevin Cocco — Sun, 08 Jan 2012 23:39:44 GMT

I highly recommend LucidCharts for flow charts and simple graphics, very slick! The word map cloud was created with the the classic Wordle. The above is a draft graphic for a research project using CrowdFlower to build a training dataset for Google Prediction API's cloud machine learning systems. Looking forward to crunching marketing data in my next modeling project!

Ice

Kevin Cocco — Tue, 13 Dec 2011 23:33:09 GMT

Alex Lemieux and I climbing GWI

Tree exit, step over, wild! Climbing up no problem, coming down hike/cartwheel ouch!

200 Terabyte Home Movie

Kevin Cocco — Mon, 12 Dec 2011 00:44:22 GMT

Some amazing graphical representations of data! Here is the company that was started from this work: Bluefin Labs The days of Neilson television-metering boxes is over. My latest projet is in the world of machine learning w/ Google Prediction API ... currently modeling a large dataset of CrowdFlower labeled tweets, fun with data, more to come ...

RIP Jamie Pierre

Kevin Cocco — Wed, 30 Nov 2011 22:29:00 GMT

Jamie was a friend of a friend and I talked with him a few times, super nice guy, sorry to see him go. More about his accident.