Selecting subsets of newly extracted features from PCA and PLS in microarray data analysis
2008

Feature Selection in Microarray Data Analysis

Sample size: 60 publication Evidence: moderate

Author Information

Author(s): Li Guo-Zheng, Bu Hua-Long, Yang Mary Qu, Zeng Xue-Qiang, Yang Jack Y

Primary Institution: Tongji University

Hypothesis

Not all top features extracted by PCA or PLS are useful, and feature selection should be performed to improve classifier performance.

Conclusion

Feature selection after feature extraction improves the generalization performance of classifiers in tumor classification.

Supporting Evidence

  • The proposed framework effectively reduces classification error rates.
  • Feature selection improves the performance of classifiers compared to using all features.
  • Not all top components are necessary for classification; tail components also provide valuable information.

Takeaway

This study shows that when analyzing gene data, we shouldn't just pick the top features; we need to choose the best ones to help our models work better.

Methodology

The study used PCA and PLS for feature extraction, followed by genetic algorithms for feature selection, and classifiers like SVM and kNN for evaluation.

Limitations

The study does not specify the limitations.

Participant Demographics

The study involved various tumor types with gene expression data from different tissue samples.

Digital Object Identifier (DOI)

10.1186/1471-2164-9-S24

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication