Feature selection and extraction strategies for non-targeted analysis using GC-MS and GC-IMS: A tutorial.

Area of Science:

• Analytical Chemistry
• Chemometrics
• Data Science

Background:

• Gas chromatography-mass spectrometry (GC-MS) and gas chromatography-ion mobility spectrometry (GC-IMS) are vital for non-targeted analysis (NTA) of volatile compounds.
• Raw data from these techniques contain numerous features, necessitating dimensionality reduction for effective chemometric analysis.
• Existing feature engineering strategies for GC-MS and GC-IMS data present challenges for researchers in selecting appropriate methods.

Purpose of the Study:

• To present and compare feature selection and feature extraction techniques for GC-MS and GC-IMS data in NTA.
• To evaluate the performance of different strategies in pattern recognition problems using these techniques.
• To provide researchers with accessible workflows and examples for applying these methods.

Main Methods:

• Review and comparison of various feature selection and feature extraction strategies.
• Application of chemometric analysis to GC-MS and GC-IMS data.
• Demonstration of practical workflows using available tools.

Main Results:

• The study highlights the importance of feature reduction in GC-MS and GC-IMS data analysis to mitigate overfitting and model complexity.
• Performance comparison of different feature engineering strategies in NTA contexts.
• Discussion on the impact of feature selection and extraction on pattern recognition outcomes.

Conclusions:

• Effective feature engineering is crucial for successful NTA using GC-MS and GC-IMS.
• This tutorial provides a comparative overview and practical guidance for researchers.
• The presented workflows aim to facilitate the application of advanced data analysis techniques in various scientific fields.