President & Founder, Elder
Research, Inc.
John F. Elder IV, PhD
John leads the US's leading data mining consulting team. Elder Research has offices in Charlottesville Virginia, Baltimore Maryland, and Washington DC.
For 20 years, the team has applied advanced analytics to achieve high ROI for investment, commercial and security clients in fields from text mining and stock selection, to credit scoring and fraud detection.
Opening Talk
In his opening keynote talk, John will reveal how data and predictive analytics are transforming business in a number of surprising areas.
John Elder Workshop
John is also running a full day workshop on Thursday (17th September) in which he will share his (often humorous) stories from real-world applications, highlighting the Top 10 common, but deadly, mistakes. Come learn how to avoid these pitfalls by laughing (or gasping) at stories of barely averted disaster.
John Elder on Target Shuffling
It's always possible to get lucky (or unlucky). When you mine data and find something, is it real, or chance? The central question in statistics is "How likely could this result have occurred by chance?"
Modern predictive analytic algorithms are hypothesis-generating machines, capable of testing millions of "ideas". The best result stumbled upon in its vast search has a chance of being spurious. Such overfit is particularly dangerous, as it leads one to rely on a model molded to the data noise as well as signal, which usually is worse on new data than no model at all. The problem is so widespread that it is the chief reason for a crisis in experimental science, where most journal results have been discovered to resist replication; that is, to be wrong!
The good news is an antidote exists! Dr. Elder will explain the simple breakthrough solution -- still rarely employed, though newly being re-discovered in leading fields. John will illustrate how to use the resampling method he calls "Target Shuffling" in multiple learning scenarios, from model fitting to data exploration, showing how it calibrates results so they are reliable - essentially providing an honest "placebo effect" against which to test a new treatment (finding).
Bottom line: Honest Data Science can save Experimental Science!
John's Background
John has Engineering degrees from Rice and the University of Virginia, where he's an adjunct professor. He's authored innovative tools, is a popular keynote speaker, and has chaired International Analytics conferences. Dr. Elder served 5 years on a panel appointed by President Bush to guide technology for National Security. He has co-authored three books (on data mining, ensemble modeling, and text mining), two of which won "book of the year" awards.