Table 2 List of classes and input parameters used for implementing the model.

From: Comparative performance of multiple ensemble learning models for preoperative prediction of tumor deposits in rectal cancer based on MR imaging

Classifier

Classes and input parameters

Weights (%)

LR

C = 0.2, max_iter = 300, penalty = ‘l2’, solver = ‘liblinear’

18.2

XGBoost

gamma = 0.003, learning_rate = 0.1, max_depth = 1, min_child_weight = 0.01, n_estimators = 50, reg_lambda = 0.001, reg_alpha = 0.1, subsample = 0.8

27.3

SVM

C = 0.2, gamma = 0.1, kernel = ‘linear’

36.4

CatBoost

depth = 4, iterations = 60, learning_rate = 0.001, l2_leaf_reg = 1

9.1

MLP

alpha = 0.15, hidden_layer_sizes = (10,10), solver = ‘adam’, activation = ‘relu’, learning_rate = 0.01

9.1

  1. LR logistic regression, XGBoost extreme gradient boosting, SVM support vector machine with the linear kernel, CatBoost categorical boosting, MLP multilayer perceptron.
Back to article page