Authors: Adeyemi Adegbenjo, Manickavasagan Annamalai
Published in: CSBE-SCGAB Technical Conferences » 5th CIGR and AGM Quebec City 2021 » Regular Sessions
Download RAW file:
Description: Spectra and imaging techniques continue to gain great attention for non-destructive food quality and safety analysis. Nonetheless, data sets often consist of attributes with varying magnitudes and scales. This occurrence was known to impede performance of learning algorithms and thereby affecting modelling accuracy. The aim of this study was to investigate the effectiveness of various normalization and scaling techniques during food quality data analysis. Various techniques considered include mean, maximum, area, logarithmic and cube root normalization, auto scaling, pareto scaling, and range scaling. The results show that the choice of normalization and data scaling technique has effect on modelling performance. In the specific case of egg strength spectra data, a combination of log-normalization and auto scaling were found appropriate and recommended preferable over other methods.
Conference name: 5th CIGR International Conference and CSBE-SCGAB AGM 2021, Quebec City,QC, 11-14 May 2021.
Session name: Monitoring, Control and Data Analysis 1
Publication type: Presentation
Language 1: en
Rights: Canadian Society for Bioengineering