Strategies and tools for variable selection in spectral profiles that combines interpretation with maximum predictive performance


Tuesday 14th November 2017

WEBINAR ON

Strategies and tools for variable selection in spectral profiles that combines interpretation with maximum predictive performance

 

14/11/17 AT 3PM (UK TIME)

 

 

Olav M. Kvalheim1, Reidar Arneberg2 and Bjørn Grung1

 

1Department of Chemistry, University of Bergen, Bergen, Norway

2Pattern Recognition Systems AS, Bergen, Norway

 

Abstract

This webinar discusses various strategies available for model interpretation and variable selection in spectral profiles. Latent variable (LV) modelling represents a key technology to obtain information and predictive models for continuous monitoring of composition and other properties of manufacturing processes.

 

In this webinar, we discuss strategies to extract knowledge and reduce LV models in a way that meet both the requirements of interpretation and optimal prediction performance for models. In particular, we focus on the problem of orthogonal variation for variable selection and strategies and tools to handle this problem.

 

This webinar will last no longer than one hour.

 

If you wish to register please contact Christine Stevenson at Christine.stevenson@strath.ac.uk

 

This webinar was previously scheduled for 18/05/17.