2025 Spring Meeting and 21st Global Congress on Process Safety
(108b) Industrial Applications of Functional Data Analysis: A Deep Dive
Functional Data Analysis (FDA) is an emerging area in statistics that offers a comprehensive way to analyze data represented as functions or curves measured over a continuous domain. In the materials and chemical industries, there are many examples of functional data - distributional data, growth/decay curves, and time series, to name a few. This has led to an increase in the use of FDA methods recently. In this talk, we present two simulated case studies, motivated by real applications at Dow. The first case study illustrates statistical testing to determine if a new material is statistically different from the current commercial offering using the FDA analogues to boxplots, ANOVA, and regression. The second showcases polymer structure-property modeling for PE blown films using a form of functional regression. Each example will outline the current modeling approaches using traditional univariate and multivariate statistical methods, discuss their performance or interpretability limitations, and highlight the benefits of using FDA methods.