In science and engineering, empirical models built with Design of Experiments and regression analysis have long helped experts predict performance in multivariate systems. 

Yet, when applied to future predictions, these models sometimes fall short. Shifts in the process, large prediction errors, or bias in the model can all cause results to miss the mark.

Today, new methods in machine learning for small experimental data offer practical ways to mitigate these issues and achieve more reliable predictions. By combining Self-Validating Ensemble Models (SVEM) with Space Filling Designs, it’s possible to reduce bias and precision errors and strengthen the accuracy of empirical models.

In this webcast, Predictum’s Cy Wegman, will demonstrate how science and engineering professionals can use SVEM and Space Filling Designs to obtain higher accuracy predictions than one might find via traditional approaches.

On completion, you’ll learn how these methods can improve predictive performance and support better decision-making in complex scientific and engineering systems.

About the presenter: Cy Wegman is a Sr. Analyst at Predictum Inc., with over 30 years applied analytics expertise across science and engineering-based industries. He is a member of P&G’s Prism Society, representing the top 2% of all P&G engineers, past and present. 

Register today to save your webcast seat!

Mitigating Prediction Errors in Empirical Models​ with SVEM and Space Filling Designs

Mitigating Prediction Errors in Empirical Models with SVEM and Space Filling Designs