Journal of Healthcare Engineering / 2022 / Article / Tab 9 / Research Article
A Comprehensive Survey on the Progress, Process, and Challenges of Lung Cancer Detection and Classification Table 9 Advantages and disadvantages of feature selection methods.
Algorithms Advantages Disadvantages GA [292 ] Tries to avoid becoming stuck in a local optimal solution GA does not guarantee an optimal solution and has high computational cost mRMR [293 ] Effectively reduces the redundant features while keeping the relevant features Mutual information is incompatible with continuous data LASSO [294 ] Very accurate prediction, reduces overfitting, and improves model interpretability In terms of independent risk factors, the regression coefficients may not be consistently interpretable SFFS [295 ] Reduces the number of nesting issues and unnecessary features Difficult to detect all subsets PCA [296 ] Selects a number of important individuals from all the feature components, reduces the dimensionality of the original samples, and improves the classification accuracy Only considers the linear relationships and interaction between variables at a higher level WONN-MLB [288 ] Integrates the maximum relevancy and minimum redundancy Has certain amount of irrelevant attributes HSOGR [90 ] Effectively selects optimized features Its execution is complex