Tyranny of Coincidence; When Spurious Relationships In Data Appear Significant

October 15, 2020 - Online

This webinar will be hosted by our partner Elder Research.

The push for organizations to incorporate analytics can result in rushed analyses with unreliable insights, even when many established best practices have been followed. A common root cause is what we refer to as the Tyranny of Coincidence, where spurious relationships in data appear significant.

Do you have data with hundreds or even thousands of possible predictors? Is your organization relying on categorical attributes with dozens of levels to inform critical models? Have you been tempted to use a “grid search” to systematically test hundreds of model parameterizations to optimally tune a model?

We will talk about how each of these scenarios lead to dangerous pitfalls without rigorous analytics discipline. At Elder Research we make model (and insight) validation a primary focus of every project, ensuring our clients can be confident that our work will hold up to time and scrutiny. In this webinar we will discuss some of the techniques we count on to avoid the Tyranny of Coincidence - target shuffling, robust combinations of cross-validation and hold out data, feature engineering, and more - with practical examples in KNIME.


How do I join the webinar?

You’ll receive a zoom link with your registration confirmation. Make sure you have a stable internet connection!

Will I be able to ask questions?

Absolutely - fire away!

Where do I find the latest version of KNIME Analytics Platform?

Download the latest free, open source version of knime here: knime.com/download

What other resources will help me to get started in KNIME?
You might also like Show all events