Sign in   |   Register

Predictive Analytics Now Easier but Still No Cake Walk

  |     |   Bookmark    
Posted February 2, 2016 By Drew Robb     Feedback

Predictive analytics is no longer the sole purview of data scientists.

While predictive analytics used to be the sole purview of data scientists and a few highly skilled analysts, it is gradually becoming more accessible to mainstream users.

Vendors are finding ways to instantly expose insights into software applications and business processes through application programming interface (API) calls, Web services, predictive model markup language (PMMLs) and other methods that integrate analysis-derived predictions with ongoing business activities in a way managers and key staff can comprehend.

"Predictive analytics vendors are providing tools that lower the barrier to entry and increase appeal for those with less statistics skills," said Mike Gualtieri of Forrester Research. "Predictive analytics has never been more relevant, and easier, than it is now."

The predictive analytics movement has benefited from the rise of Big Data, Hadoop and modern agile software development tools. Combined with almost unlimited compute power, the latest predictive models are more efficient, accurate and available than ever.

In this article, we cover:

  • Some ways predictive analytics can benefit business
  • How machine learning is impacting predictive analytics
  • Some tips on successfully implementing predictive analytics

How Predictive Analytics Benefits Business

Used correctly, predictive algorithms are good at detecting patterns in data that can then be extrapolated to predict similar outcomes in the future. For instance, telecommunications and other firms may notice certain behaviors in common with those who desert their services. As a result, they can take preventive steps to retain those that fall into the danger zone.

A predictive analytics model that was built to identify customers with a high risk of churn can be integrated into a call center application to notify call center agents of the risk and provide possible actions when the customer calls in.

"This integration of predictive analytics to support or even automate decision making in organizations is a key trend that we are seeing in the market," said Sascha Schubert, director, Analytics Product Marketing for SAS, which IDC tapped as the predictive analytics leader with a 33 percent share of the market in 2014.

Predictive Analytics and Machine Learning

Looking further down the road, it may well be that the human element is eliminated from much of the work of predictive analytics. Jeff Erhardt, CEO of predictive analytics provider, sees the evolution of machine learning taking on far greater importance. While classical statistical analysis relies on a human expert to formulate and test the relationship between cause and effect, machine learning eliminates the human bottleneck.

"Machine learning is an important emerging technique that is enabling the next generation of predictive analytics use cases," Erhardt said.

Ultimately advanced software and machine learning could take the analyst and data scientist out of the broad business of business intelligence and analytics. Only when anomalies or exceptions are isolated, would the human element be entered into the equation.

A good analogy is air travel. While the pilot is little more than a spectator for most of the journey, his or her presence remains essential to overall success. Auto-pilot, machine learning and software-driven predictions will get us so far. But human judgment will never entirely disappear. 

A Platform Approach to Predictive Analytics

Many firms are likely to have some kind of business intelligence or analytics applications already running. So when and how should they add predictive analytics to the mix?

"Most people think about adding predictive capabilities to their existing stack via a standalone tool," said Bruce Kolodziej, Predictive Analytics Sales Manager at Information Builders, a provider of analytics software. "A better way is to think about it more broadly, as predictive applications require things working in concert: data access and preparation, predictive model building and testing, then operationalizing the results for business end-users."

He advocates a platform approach where everything works together. Without this, he said, organizations end up with siloed tools that are not integrated.

Predictive Analytics Is Not Easy

That said, Steven Hillion, co-founder and chief product officer of advanced analytics provider Alpine Data Labs, cautions organizations not to expect instant miracles from predictive analytics.

"While there's a great interest and excitement in predictive analytics today, it seems like success stories are the exception rather than the rule," he said.

While many organizations would love to emulate the data-driven prowess displayed by leaders like Amazon, Google and Netflix, it is not easy to achieve tangible, repeatable business results.

A fundamental reason for this, Hillion believes, is the inherent complexity of predictive analytics projects. It is easy to become overwhelmed with technical details around cleansing data and other areas. It takes a business-driven approach to ensure success.

"Amazon and Netflix are able to use data in innovative ways not just because they are technically advanced, but also because they've created a culture of analytics that pervades every aspect of their business," he said.

Get Business Involved in Predictive Analytics

Emulating these data darlings is no easy task. Doing so means wisely integrating predictive analytics into the fabric of the business. Before jumping into the deep end with highly complex technologies and advanced algorithms, companies should address low-hanging fruit and build analytic applications that are valuable to actual business users, advised Hillion.

In other words: Don't let the perfect database or latest and greatest statistical model get in the way of achievable results.

In addition, Hillion said, the best way to integrate predictive applications with business intelligence is to first identify the areas in which they are already being used. In some cases, the predictive analytics use case may warrant an entirely new user interface, while the BI application can still work at tasks like clustering historical data or visualizing audience demographics.

That requires business experience and insight.

Ready to get started? Check out our five-step guide for kicking off a predictive analytics project; our nine tips for data preparation for analytics; and our buying guide for predictive analytics software.

Drew Robb is a freelance writer specializing in technology and engineering. Currently living in Florida, he is originally from Scotland, where he received a degree in geology and geography from the University of Strathclyde. He is the author of Server Disk Management in a Windows Environment (CRC Press).

Page 1 of 2   |   1 2   |   Next

Submit a Comment

Loading Comments...

Thanks for your registration, follow us on our social networks to keep up-to-date