Subscribe to DSC Newsletter

Sorry I am late for commenting in the Gartner Hype Cycle for Emerging Technology 2014 that was released back in August.

I was prepping for a presentation recently and thought that referring back to the Hype Cycle published by Gartner would be an interesting way to talk about some of the trendy terms in the world of Advanced (Predictive and Prescriptive) Analytics.   I was looking for some of the overused terms like Data Science, Big Data, In-Memory, Predictive Analytics, and Prescriptive Analytics.  Shocker!  Predictive Analytics doesn’t appear on the list anymore!

Looking at the 2013 version of the Hype Cycle, I see that Predictive Analytics reached the “Plateau of Productivity” in 2013.  How could this be when so many companies that I talk to – substantial Fortune 500 companies – are not yet operationalizing predictive analytics? 

Does Gartner have it wrong?  Or are my expectations incorrect?  Gartner defines Predictive Analytics as:

Any approach to data mining with four attributes:

  1. An emphasis on prediction (rather than description, classification or clustering)

  2. Rapid analysis measured in hours or days (rather than the stereotypical months of traditional data mining)

  3. An emphasis on the business relevance of the resulting insights (no ivory tower analyses)

  4. (increasingly) An emphasis on ease of use, thus making the tools accessible to business users.

Based on Gartner’s  definition, any organization that is using Excel to develop a prediction or forecast, regardless of the ability of operationalize it, is doing Predictive Analytics!

Unfortunately, until organizations take the results of the Predictive Analytics and embed it into operational processes, the value of the prediction will always be muted:

  • What good is it for me to predict how many customers will churn next month unless I can do something to save them?

  • How beneficial is it for me to identify potential fraud unless I can put a prioritized list (based on the ability to investigate, for example) into a case management system, along with the relevant case information so that the cases can be investigated expeditiously?

  • What good does it do me to forecast demand for a product, unless I can adjust my supply chain, and pricing to maximize revenue and meet demand?

  • Should I mail to everyone that I predicted would buy?  Or just those that my budget allows?  Or just those that I can have the greatest impact over?

  • What value does it provide me to know that a patient will be readmitted to the hospital unless I can put processes in place that will keep them from being readmitted?

The Bottom Line: The true value of Advanced Analytics can only be obtained when the results of the predictions are layered with business rules, optimized, and embedded into processes at the point of decision-making.   

A report doesn’t cut it.

To get the full value out of Predictive Analytics, you need to be doing Prescriptive Analytics.  Unfortunately, according Gartner’s Hype Cycle, Prescriptive Analytics is only at the “Innovation Trigger” phase at this time.

Laura E. Wood Squier is director at Quebit, a leading provider of IBM SPSS Advanced (Predictive and Prescriptive) Analytics Services.  As the first IBM SPSS Partner to obtain Gold Accreditation on IBM SPSS Modeler in North America, QueBit's team team brings many years of hands-on experience implementing Prescriptive Analytics Solutions.    

Views: 1787


You need to be a member of AnalyticBridge to add comments!

Join AnalyticBridge

Comment by Ralph Winters on October 17, 2014 at 2:16pm

I agree with you that predictive analytics is gaining momentum, rather the losing it. It's very possible that some terms drop off the right side of the cycle, only to be 'reborn' with a new name, as new terms appear on the left. It keeps things novel and fresh!

Comment by Laura E. Wood Squier on October 16, 2014 at 11:29am

I agree Angus.  We also have to get people out of the mindset that a model is perfect.  It is never going to be that way.  Rather than assessing a model on overall goodness of fit, people need to assess on business value.  the model won't be perfect, but it is a &377! of a lot better than guessing!

Comment by Angus Urquhart on October 16, 2014 at 10:56am

Great article.  I work in a consultancy that works with companies who are trying to get more from their data and I'd say that the majority are barely scratching the surface in terms of predictive analytics.  

I think this is partly because they don't feel their data is ready, partly because they don't have the right internal skills and partly because they've not managed to get the value from predictive analytics in the past.  

The first 2 reasons I think can be fixed fairly easily by spending some time working with your data and getting better people.  The 3rd is more about realising how to use predictive analytics correctly.  For example, someone might say that sales are so volatile that only sales reps can predict sales and that no algorithm will do a better job.  Well that can be true but you can then use predictive analytics to strip out human bias in their forecasts.  I think we'd see a lot more adoption if people figured out the best use cases for this type of analytics.

On Data Science Central

© 2020 is a subsidiary and dedicated channel of Data Science Central LLC   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service