Reduction of repeatability error for analysis of variance-Simultaneous Component Analysis (REP-ASCA): Application to...

Area of Science:

• Chemometrics
• Analytical Chemistry
• Data Analysis

Background:

• Repeatability error in multivariate data can significantly distort results from techniques like Analysis of Variance-Simultaneous Component Analysis (ASCA).
• Physical variations during data acquisition, such as in Near-Infrared (NIR) spectroscopy, introduce baseline shifts that obscure true chemical information.

Purpose of the Study:

• To develop and evaluate a novel method, Repeatability Error in Multivariate Data for Analysis of Variance-Simultaneous Component Analysis (REP-ASCA), to mitigate repeatability errors.
• To improve the reliability and interpretability of ASCA when applied to datasets with significant measurement variability.

Main Methods:

• The REP-ASCA method involves adapting the acquisition protocol to include repeated measures for error estimation.
• Orthogonal projection in the row-space is used to reduce the identified repeatability error from the original dataset.
• ASCA is subsequently performed on the orthogonalized dataset to analyze the corrected data.

Main Results:

• The developed REP-ASCA method effectively reduces repeatability error in multivariate data.
• Application to NIR spectral data of coffee beans demonstrated that REP-ASCA removes baseline artifacts caused by physical variations.
• Factor analysis on the REP-ASCA processed data provided clearer insights into the chemical composition related to the factors of interest.

Conclusions:

• Repeatability error is a critical factor that can significantly impact the outcomes of variance-based data analysis.
• The REP-ASCA method offers a robust approach to correct for such errors, thereby enhancing the chemical interpretability of spectral data.
• This technique is particularly valuable for analyzing complex samples where measurement consistency is a challenge.