D19 - Deploying DOE to Predict Process Performance
Trish Borzon
1370 Posts
Design of experiments (DOE) is a tried-and-true, multi-factor quality tool for identifying key process drivers. This presentation demonstrates how to deploy DOE to create reliable prediction models. Similar in concept to estimating the power of a design, prediction precision becomes the key evaluation statistic. A case study demonstrates how to confirm that a particular design will provide the desired results, more reliable process settings. Although DOE tools require a high level of stats and math, modern-day software bears the burden of the computations. Therefore, the focus of this talk will be kept to the need-to-know elements for successful experiment design, analysis, modeling and optimization. Ultimately, effective application of DOE produces very useful predictive models that lead to profound process understanding. Participants will be briefed on the key statistics that quantify the ability of the model to make good predictions, as well as tools for final optimization.
1 Replies
Thank you for allowing me to present on my favorite topic - how to get the most out of your design of experiments! If anyone has questions, they are welcome to email me via shari@statease.com.
Shari Kraber
Go back to WCQI Home Page