‘Big data’ as a buzzword has captured just as much attention as any.
For the most part this attention is well deserved, but I’m afraid that in paying so much attention to big data we may be leaving out the most important part: analytics. That is to say, creating, documenting, and retrieving vast amounts of data is one thing, but understanding it is a whole other issue.
First let’s start with how we got to where we are now.
Data started to get ‘big’ some time ago. With the digitization of our daily lives, the ability to quantify just about every interaction became a very real reality. Think of the sheer volume of data being collected by Google searches and transaction logs by the likes of Wal-Mart or Amazon.
Organizations in this sphere recognized the value to be mined from their big data early on and have been ahead of the curve to make the transition to ‘big data analytics’.
But the value proposition associated with analyzing newer data streams has had less time to take hold. Now it seems obvious that there is value in organizations being able to analyze vast streams of social media data and compiling profiles to better target customers. However, fully delivering on this notion still has some distance to cover.
Consider the challenge that comes with analyzing the increasingly amorphous state of data. Performing sentiment analysis of your brand might be straightforward in so far as measuring ‘likes’ or searching for specific keywords and phrases, but the endeavor is considerably harder if you want to understand pictures or videos.
Our next challenge will come with the rise of the ‘Internet of Things’ and the accompanying proliferation of machine sensor logs. Machine data has already been used for some time.
Sensors on things like wind turbines and manufacturing lines have long ago proven their worth, but what if you extend the concept to consumer devices as well? Every phone, every household appliance, and every piece of wearable technology has the opportunity to produce countless points of data in just about every conceivable form.
So the question remains, how will we analyze this all?
Demand for Big Data Analytics
In the process of transitioning from big data to big data analytics there is a challenge of making it all relevant to the business user. Organizations recognize that they need analytics, but they don’t want to have to hire a team of PhDs and data scientists to build complex models and perform involved statistical analysis every time they have a question.
To this end we have seen the rise of a number of players who specialize in getting to big data analytics in an attainable way.
Companies like Sisense or Platfora have the chance to provide us with direct access to the Hadoop file system in our quest to achieve real-time analysis. And still others such as HP Vertica allow for organizations to plug into existing legacy data warehouses. This is an important piece of the puzzle to solve, as one of the greatest promises of big data analytics is to get better insight from the data that we have already collected.
A lot of noise has been made around discovering the ‘dark data’ within organizations. With good reason too: it’s tantalizing to think of extracting value from the years and years of logged data that have – until recently – needed to be buried away in the dark depths of organizations.
In either case, the future of big data analytics will likely involve an increasing amount of machine learning.
What It Means
Data sets can be too complex and unwieldy for us to analyze in an efficient manner without assistance. Advanced algorithms and pattern recognition help scale massively complex data sets to a level that can be manipulated in a meaningful way.
Think of it like power steering for your car: we need people to drive the analysis, but a bit of mechanical help makes it a whole lot easier. A host of companies are rising to the occasion in this arena as well, although an interesting one to watch is Ayasdi, particularly with their applications in pharmaceutical drug developments.
As we continue to sprint headlong into a future of creating more and more data, the need to analyze it in a useful way will continue to compound. The market is responding in kind by producing a myriad of solutions to tackle our mountains of data.
Whether it is from finally leveraging data we have long been collecting or a newfound ability to make sense of previously incomprehensible data sets, the future of big data lies within analyzing it. Businesses will do well to consider the value which they can potentially achieve when they make the transition from just big data to big data analytics.
About the Author: James Haight is a research analyst at Blue Hill Research focusing on analytics and covering a variety of emerging enterprise technologies. James brings a unique perspective to technology research, drawing from his background in economics, executive compensation, and microfinance.