Aberdeen Group - Home
Predicting Faster Predictive Analytics Modeling
David White
MAY 14, 2013 09:14 AM
A+ A A-

Predictive analytics is moving mainstream — big data is making it so. Aberdeen’s big data research ("Go Big or Go Home? Maximizing the Value of Analytics and Big Data") indicates that 90% of organizations working with big data were also using predictive analytics. This is a natural fit — as data sets get bigger and more complicated then the human eyeball is going to find it much harder to find patterns in data. Time for smart software in the form of predictive analytics and data mining technology to step in and help.

But as predictive analytics becomes more widespread, so the imperative to build and refine models more efficiently will grow. Upcoming research from Aberdeen has some answers to that problem. The research will differentiate Leaders (the top performing 35%) from Followers (the other 65%) using two measures:

  • The average time required to build and deploy a new predictive model once the requirements have been recognized.
  • The average time required to refresh an existing predictive model, when necessary.

Leaders are able to build new models in nearly half the time taken by Followers, and are able to refresh models over ten times faster. One of the reasons for this is that Leaders are much more likely to have automated support for key steps in the predictive analytics development lifecycle.

Figure 1.

figure 1

As Figure 1 shows, while data preparation was mostly or fully automated at 53% of Leaders, only 38% of Followers could make that claim. This is particularly important, since 28% of the total time required to implement and deploy a new predictive model is taken up by data preparation, on average. The same trend — higher use of automation —l was also apparent for both model building and model deployment. Simply put, automation can eliminate slow, error-prone manual tasks, hence accelerating the delivery of new predictive models.

This and other findings are due to publish in a new research report at the end of May 2013.

In the meantime, you can get free access to my earlier work on predictive analytics (and thousands of other research reports) by registering for AberdeenAccess.

by David White
Senior Research Analyst
Business Intelligence (BI)
Aberdeen Group

[%= name %]
[%= createDate %]
[%= comment %]
Share this:
Please login to enter a comment: