This material was presented at the 8/29/18 Novozymes event in Cary, NC at the SAS EBC. It is an example of using Functional Data Explorer on a seven function fermentation process with sensors on Ethanol, Temp, Molasses Feed, NH3 Feed, Air, Tank Level and pH. Data was collected over 100 batches with 400 sensor readings taken at regular intervals from start to finish. The PowerPoint presenation walks through a 7-step work flow that results in a generalized regression model for variable selection which answers the question, which principal component waveforms are showing a significant influence on the response? The response in this case is the final yield of ethanol produced. There is an "extra credit" 8th step that uses Bootstrap Forest on a wide (split) version of the data to determine which snippets of the significant principal component waveforms are making the strongest contribution.
