Self-expansion full information optimization strategy: Convenient and efficient method for near infrared spectrum...

Area of Science:

• Chemometrics
• Spectroscopy
• Data Analysis

Background:

• Spectrum preprocessing is crucial for near-infrared (NIR) spectroscopy accuracy.
• Manual preprocessing methods are time-consuming and prone to errors, especially for less experienced users.
• Existing automated methods have limitations in determining preprocessing sequences and providing statistical information.

Purpose of the Study:

• To develop an automated spectrum analysis methodology for NIR data.
• To address challenges in preprocessing sequence determination, optimization outcome fluctuations, and statistical information.
• To establish a reliable and effective automatic near-infrared auto-modelling method.

Main Methods:

• Introduction of the self-expansion full information optimization strategy.
• Utilizing built-in modules for information generation and spectrum processing.
• Employing Monte Carlo cross-validation to search preprocessing techniques and wavelength selection algorithms.
• Building ensemble calibration models with optimized preprocessing schemes.

Main Results:

• The proposed method successfully extracts spectral information and develops calibration models.
• Demonstrated effectiveness on two sets of real-world NIR spectral data.
• The strategy provides objective statistical information for model evaluation.

Conclusions:

• The self-expansion full information optimization strategy offers a powerful, open-source solution for NIR spectrum auto-analysis.
• It enables concurrent resolution of preprocessing sequence, optimization, and statistical information issues.
• The methodology shows broad applicability to other spectroscopic techniques like Raman and infrared spectroscopy.