Earlier this week I posted on the value of decision requirements modeling in analytic projects when it comes to coping with some of the analytic skills shortages people face. But this is not the only reason to focus on decision requirements if you are focused on predictive analytics and data mining. In fact decision requirements modeling has a role in the analytic lifecycle more generally.
Take this SAS white paper as an example – Manage the Analytical Life Cycle for Continuous Innovation – From Data to Decision. This lays out a nice (and fairly typical) sequence:
- Problem identification
- Data preparation
- Model Development
- Model Validation
- Model Deployment
- Monitoring and assessment
The paper also (correctly) identifies that it is critical that staff from different backgrounds (business, IT, analytics – what I call the three legged stool of successful analytics) are involved. However like every analytic tool vendor out there SAS then begins by talking about how their software tools can help with everything from data preparation and exploration to model monitoring and assessment. But what about problem identification?
It is in problem identification that decision requirements modeling really pays off for analytic projects. Decision requirements modeling provides the formal tools and techniques you need to develop business understanding for analytic projects. Established analytic approaches such as CRISP-DM as well as all the major analytic tools vendors stress the importance of understanding the project requirements from a business perspective. While most organizations officially take this position too, the reality is that most do not have a well defined approach to capturing this understanding in a repeatable, understandable format. Decision requirements modeling closes this gap and develops a richer, more complete business understanding right at the start of an analytic project. Specifically decision requirements modeling gives you:
- A clear business target defined in terms of KPIs/metrics to be influenced
- A precise definition of where in the decision-making the analytics will have an impact
- An understanding of how the results of your analytics will be used and deployed, and by whom
As noted earlier it also reduces reliance on constrained specialist resources by improving requirements gathering and i
- mproves collaboration across the organization. If your analytic projects struggle to be deployed or used, or thrash around trying to determine exactly what the analytic is for, why not d
ownload the paper to learn how to do decision requirements modeling for analytic projects.