Using Kernel Methods for Identifying Tobacco Varieties
Author Information
Author(s): Martin Florian
Primary Institution: Philip Morris International R&D
Hypothesis
Can kernel methods effectively encode SSR marker polymorphisms for the identification of tobacco varieties?
Conclusion
The proposed method allows for accurate and economical identification models based on SSR genotyping.
Supporting Evidence
- The method outperformed other filter methods in selecting SSR markers.
- The approach led to satisfactory prediction models when combined with kernel linear discriminant analysis.
- The study demonstrated that fewer than 8 markers can achieve similar classification performance.
Takeaway
This study shows a way to use genetic markers to tell different types of tobacco apart, which can save time and money.
Methodology
The study used kernel methods to encode SSR polymorphisms and a feature selection algorithm to build prediction models.
Limitations
The study suggests that using fewer markers can lead to weaker predictions if new genotypes are not represented in the original dataset.
Participant Demographics
The study involved various tobacco varieties, including Burley, Flue Cured, and Oriental types.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website