Table of contents for Predictive Analytics World 2009
- 5 ways to reduce cost with predictive analytics
- SAS and the art and science of better
- The High ROI of Data Mining for Innovative Organizations
- High-Performance Scoring of Healthcare Data
- Completing the visitor targeting cycle
- New Challenges for creating predictive analytic models
- Predictive modeling and today’s growing data challenges
- The unrealized power of data
- Expert Panel on Challenges and Solutions
- Predictive Modeling for E-Mail Marketing
- Analyzing and predicting user satisfaction with sponsored search
- Some thoughts after attending Predictive Analytics World
Syndicated from Smart Data Collective
Anne Milley from SAS, one of the sponsors of the show, spoke on the art and science of better. Data is often messy and the enterprise is not a lab. Nevertheless, she says, we can still bring science to bear. We can observe, define, measure, experiment, learn and ACT. Anne had a number of observations:
- We must begin with observation. Semmelweis’ study on hand washing in 1847 observed that hand washing saved lives but without the understanding of these observations nothing could be done.
- Defining the right problem is essential. For instance, in CRM the results are very different if time is considered (e.g. with survival methods) than if it is not.
- While there is a cost of running experiments to see what you can learn, there is a cost of ignorance too. Collecting more data through experiments may cost money but not knowing can be much more expensive.
- Creating a culture of experimentation and continuous learning is both essential and difficult.
- There is an essential step of acting on the modelling or analysis. As Deming said “The object of taking data is to provide a basis for action“. This often requires more interpretation and discussion than might be expected and tools like visualization can really help explain what a model is saying, thus increasing the likelihood of action.
- Challenging business as usual is a great way to use analytics and this can be supported by developing an Analytic Center of Excellence (though I would say a Decision Center of Excellence would be better) to see what is being done best across the company, close the loop and drive new behavior elsewhere.
Anne ended by pointing out just how important the social dimension can be for actually putting analytics to work and be more impactful.