Skip to main content

Table 1 Accuracy of classifiers used in the experiments on the TCGA dataset

From: Automatic discovery of 100-miRNA signature for cancer classification using ensemble feature selection

Classifier Accuracy (10-fold CV)
  1046 Features 100 Features Hyper parameters Feature selection method
  avg std avg std   
Gradient Boosting 0.9398 0.0076 0.9359 0.0086 300 predictors Decision Trees
Random Forest 0.9351 0.0071 0.9324 0.0073 300 predictors Decision Trees
Logistic Regression 0.9178 0.0096 0.9237 0.0067 - Coefficients
Passive Aggressive 0.9117 0.0104 0.8831 0.0115 - Coefficients
SGD 0.91 0.0074 0.9035 0.0152 - Coefficients
SVC 0.9211 0.0122 0.9154 0.0065 Linear kernel Coefficients
Ridge 0.8971 0.0138 0.8305 0.0062 - Coefficients
Bagging 0.9151 0.0120 0.9110 0.0077 300 predictors Decision Trees
Average 0.918463 - 0.9044 - - -
  1. In the case a classifier is not using standard values for its hyperparameters, the relevant variations are summarized in the corresponding column