Litcius/Paper detail

Comparison of Norm-Based Feature Selection Methods on Biological Omics Data

Jiayuan Song, Zheng Liu

202115 citationsDOI

Abstract

Feature selection methods have become significant methods when analyzing high-throughput biological data due to the nature of large p and small n problems. One of the most crucial categories of feature selection methods is norm-based approaches because they can reduce the magnitude of coefficients and enhance the sparsity of selected features. There are many norm-based feature selection methods with different merits and demerits. Therefore, the specific choice of norm-based methods for omics data has become a problem. In our work, we mainly concentrate on the comparison and evaluation of two popular norm-based methods, namely Lasso and Ridge regression. The regression with norm is Lasso Regression and the regression with norm is Ridge Regression. The results indicate that Ridge Regression performs better than Lasso Regression when dealing with high throughput TCGA datasets.

Topics & Concepts

Feature selectionNorm (philosophy)Lasso (programming language)RegressionElastic net regularizationComputer scienceRegression analysisLinear regressionArtificial intelligenceData miningMachine learningMathematicsStatisticsWorld Wide WebPolitical scienceLawGene expression and cancer classificationBioinformatics and Genomic NetworksMachine Learning in Bioinformatics
Comparison of Norm-Based Feature Selection Methods on Biological Omics Data | Litcius