Predictive Analytics: Extending the Value of Your Data Warehousing Investment -- Executive Summary
January 9, 2007
This report is designed for the business or technical manager who oversees a business intelligence (BI) environment and wishes to learn the best practices and pitfalls of implementing a predictive analytics capability. The report defines predictive analytics as a form of BI that uncovers relationships and patterns, within large volumes of data, that can be used to predict future behavior and events. Unlike other BI technologies, predictive analytics is forward-looking, using past events to anticipate the future.
Predictive analytics delivers a significant return on investment compared to other BI technologies. Analytical models can optimize a wide variety of business processes in different industries and functional areas. But predictive analytics is not a silver bullet.
To create an effective analytical model, a person needs deep knowledge of a business process, strong analytical and computer skills, and a good grasp of the data and underlying systems. It takes an experienced modeler about three weeks to create an effective model from scratch using a six-step process that includes: (1) define business objectives, (2) explore the data, (3) prepare the data, (4) create analytic models, (5) deploy the models, and (6) manage the models. To support a successful predictive analytics practice, companies must invest about $1 million annually on labor, software, hardware, and services.
The advent of high-powered computing platforms, new graphically-oriented analytic workbenches, and data warehouses has dramatically changed the value equation for predictive analytics. These new technologies make it possible for business analysts rather than statisticians to create effective analytic models faster and more effectively than in the past and embed them in operational applications so the business gets immediate benefit.
Countless case studies attest that predictive analytics drives substantial business value. Yet, most companies haven't figured out how to implement or nurture a predictive analytics capability, and its statistical and mathematical origins tend to intimidate IT and business managers. To help organizations implement predictive analytics, this report provides five best practice recommendations:
- Hire business-savvy analysts to create analytical models.
- Nurture a rewarding environment to retain analytic modelers.
- Fold predictive analytics into the information management team.
- Leverage the data warehouse to prepare and score the data.
- Build awareness to build confidence in the technology and process.