Sparse PLS Discriminant Analysis for Multiclass Problems
Author Information
Author(s): Lê Cao Kim-Anh, Boitard Simon, Besse Philippe
Primary Institution: University of Queensland
Hypothesis
Can sparse PLS discriminant analysis (sPLS-DA) improve variable selection and classification performance in multiclass biological data?
Conclusion
sPLS-DA performs comparably to other methods while offering better computational efficiency and interpretability through graphical outputs.
Supporting Evidence
- sPLS-DA shows competitive classification performance on public microarray and SNP data sets.
- The method provides valuable graphical outputs for better interpretation of results.
- sPLS-DA is computationally efficient compared to other variable selection methods.
Takeaway
This study shows a new way to pick important features from complex biological data, helping scientists understand diseases better.
Methodology
The study introduces sPLS-DA, which combines variable selection and classification in one step using a sparse version of PLS.
Limitations
The study does not address the issue of unbalanced classes in multiclass problems.
Participant Demographics
The study analyzed data from various cancer types and SNPs, including samples from different populations.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website