2020 Virtual Spring Meeting and 16th GCPS
(59k) Real-Time Multivariate Data Analysis for Real People
Authors
Zink, A. - Presenter, The Dow Chemical Company
-
User acceptance of analytics tools is key to the Digital journey. Traditional univariate statistical data analyses (control charts, etc.) are relatively well accepted because non-analytics-experts can digest the results and take action. Advanced machine learning methods have the potential to alert earlier and/or on more subtle trends. However, the challenge is to provide enough information about the machine learning result for a user not trained in analytics to take action. This poster covers the deployment of Northwest Analytics' multivariate monitoring capability and an example case study. The interplay between multivariate and univariate results on one deployment demonstrates that the two approaches are quite complementary.