First analytics roundup for 2016

I made the mistake of not working between Christmas and New Year.  Too much work piled up while I was away and I didn't get to blog my first weekly analytics round-up of 2016!  So here it is...

• In my machine learning strategy report (subscribers only, sorry...) I touched on Netflix as an example of machine learning being applied in the consumer world.  For those of you that want to dig a little deeper, this would be the article:  “Because You Watched…” How Netflix Makes Recommendations.


• And here's your first of many cloud stories for 2016.  The cloud isn't going away anytime soon - in fact, I think we'll see it march relentlessly onward this year.  Amazon are generally reckoned to have a much bigger public cloud than the other leading vendors combined.  But, scale isn't everything.  Then again, it seems Amazon also beat its rivals handily in terms of uptime and availability.  With 2 1/2 hours of downtime though, that's roughly 99.983% availability.  That's still some way short of the mythical "five nines" (99.999%) availability often demanded for highly available mission-critical systems.  However, getting an extra hour or two of uptime is way more costly.  In practice, the majority of applications in most businesses can survive a bit of outage perfectly well.


• I've spoken to the team at Sisense a couple of times over the years.  I've never used the product in all honesty, but it's always struck me as a nifty, data prep-to-visualization solution in a single tool. Consequently, it's pleasing to see the company just scored an additional $50 million in VC funding.  I'm not quite sure why the article I linked to puts a bull's-eye on SAP, as Sisense seems pretty adept at dealing with a wide range of data from many sources.


• Finally, it's never too early for a bit of hype and buzzword bingo, is it…?  Datanami asks "Is 2016 the Beginning of the End for Big Data?"  Personally, I think it can't come a day too soon!  Surely the most misused term of the last few years, I can only pray that I've done more to clarify than confuse.  Frankly, many organizations still struggle to get the right information to all of the managers that need it in a timely way - regardless of the size and complexity of the data set. However, the Industrial Internet of Things (IIoT) will only add to the volume, velocity, and variety of data that industrial organizations need to manage.  So will still face the challenge of managing these data sets, I just hope we can finally get away from all the hype and nonsense associated with the term big data.


• And talking of hype and buzzwords, Datanami also has a feature on what skills data scientists need.  It seems Python is right at the top of the list - a language typically used more for data preparation than anything else.  So here we are.  A brand-new year, and yet apparently data cleansing and data preparation is still the number one issue for analytics.