On Monday, we review news and resources related to predictive analytics. This week there are five stories we think are especially noteworthy:
- The Dos and Don’ts of Predictive Analytics: This simple slideshow is good for explaining the basics of predictive analytics to people who are otherwise confused by or leery of the technicalities. The guidelines are from Rado Kotorov, chief innovation officer at business intelligence and data analytics firm Information Builders.
- VIDEO: How to Take Predictive Analytics Beyond the Manufacturing Floor: This provides an overview of how the “factory of the future” can be developed with predictive analytics. This short video has a bit of an infomercial feel at first, but Clint Belinsky, VP of global quality for Jabil Circuit, goes on to provide a good overview of how his organization uses an y=f(x) formula to predict issues before they occur. Jabil combines data from suppliers, the manufacturing process, and final assembly procedure. Then it places that data into a predictive model that alerts engineers to potential problems before they occur. It also gives operators prescriptive instructions to help them proactively fix problems.
- ADP to Offer Predictive Analytics Tool for Flight Risk of Workers: As the pool of available skilled employees shrinks, companies are trying to reduce turnover rates. Therefore, the vendors of cloud-based human capital management software and services are scrambling to create predictive analytics tools to help customers identify when employees are at risk of leaving their positions. ADP plans to release a new tool in this area. TechTarget reports, “Marc Rind, chief data scientist and vice president of product development at ADP … said in order to determine flight risk, the cloud-based tool will draw on aggregated and anonymous data from employees at 75,000 U.S. businesses. For example, one out of every six workers in the U.S. receives a payroll check from ADP.”
- Taking Action on Technical Success: A Fable of Data Science and Consequences: This is a fascinating though fictionalized account of how a company creates a model to predict customer churn and then discovers that the model’s real-world implementation has unintended consequences. The organization comes to realize that its predictive model does not have a way of deciding when to contact a customer in danger of dropping its services. Nor does the model have a way to measure its effectiveness compared to an alternate reality where the model is not applied. The article does a nice job of highlighting the importance of combining prediction with action planning.
- Improve New Product Development with Predictive Analytics: This article examines how predictive analytics can boost the odds that a new product will succeed. After all, the chances of failure are high, with Deloitte finding that 96% of product innovation fails to return the cost of capital. The authors, Tom Davenport and Andrew Spanyi, write, “The power of predictive analytics is multiplied when an organization takes an end-to-end process view of new product development (NPD).” Then they go on to discuss the steps in the NPD process.
As predictive analytics (PA) becomes more mainstream, experts are trying to attain several goals. First, they want to communicate the basics of PA so that decision makers will feel more comfortable using these techniques. Second, they are using PA to complement and strengthen a wide range of existing tools, such as statistical process control (SPC) in manufacturing, human capital management software in HR departments, and testing in new product development. Third, they are refining PA techniques in order to improve on their efficacy in real-world situations. It’s not enough to make accurate predictions. Organizations must know how to be put that knowledge to work.