Your roundup of breaking analytics news

Not good.  Friday already, and my name is on the calendar for today's blog...Lucky for me there's so much interesting analytics stuff to write about:

• IOT will drive mass adoption of platform as a service (PAAS) and radically change application design.  Well...perhaps.  The headline is stuffed full of poetic license.  But, if we can ignore that for a minute, there is some interesting stuff here.  That big analyst firm that begins with G reports that most new IoT-centric solutions will be implemented on IoT platforms.  Which I'm inclined to agree with, I just don't think it justifies the use of "mass adoption" in the headline.   (You know those Brits *only* appear to be quiet and humble...).  I think Platform-as-a-Service is a natural fit for Industrial IoT.  The cloud can connect across organizational boundaries easier, and scaling storage and processing isn't a problem either.  A truly modern PaaS is also built to facilitate micro-services which also aid agility and flexibility.  You can find the initial thoughts of Greg Gorbach and I on IIoT platforms here.  We're also working on a research strategy report on that topic at the moment, more on that later...


• And talking of all that IoT data, the latest iteration of the Raspberry Pi - the predictably named Raspberry Pi 3 - has buffed its credentials as an IoT edge device.  Now sporting a 64-bit quad-core clocked at 1.2 Ghz it should have the muscle to run analytics at the edge.  But, more importantly I think, it also has built-in wi-fi and Bluetooth support, which could be a big win with industrial users.  Learn more here, here, and here.  Oh, and remarkably it's still yours for just $35 - and I imagine you could do rather better than that if you bought in volume...


• A couple of articles published this week that served to remind us that analytics alone is never enough. For example, Andrew Pease at SAS wrote about the apparent cooling on sports analytics at the Boston Red Sox.  And Paul Goad of The Guardian (yes, more Brits...) wrote that blind faith in big data may be dangerous.  The point is that blind faith in analytics is never the right approach.  For example, if you look at the data in a chart and it doesn't show what you expected, one of two things has happened.  First, something unexpected has actually happened in your operation/business and you need to take action to address it.  Second, the data isn't actually correct - perhaps something went wrong in the collection or preparation of the data.  We don't want analytics to replace human brains, we want analytics to provide insights that human brains can make rational decisions about.


• Finally, if anyone has dared to dip their toes into the machine learning waters, this article has some nice guidance for being productive over the long haul.