Skip to main content

Table 2 The parameter settings

From: Joint modeling strategy for using electronic medical records data to build machine learning models: an example of intracerebral hemorrhage

Models

Packages

Parameters to be tuned

Parameters ranges

Optimal parameters

LR

–

–

–

–

RF

randomForest

mtry: number of randomly selected

variables

mtry = 1:9

mtry = 5

ANN

nnet

size: numbers of hidden units,

decay: weight decay

Size = 1:9,

Decay = (0, 0.1, 0.01, 5e-4)

Size = 5,

Decay = 0.01

SVM

Kernlab

sigma: Sigma*

, C: cost

Kernel = Radial basis function Kernel,

C = (0.25, 0.50, 1) **

C = 1

KNN

–

k: number of neighbors

k = (5, 7, 9) **

k = 5

Stacking

caretEnsemble

–

–

–

AdaBoost

fastAdaboost

nIter: number of trees

nIter=(10,20,50,100,150,200,300,500)

nIter = 20

  1. –: No parameter needed to be tunned; *: The optimal value was automatically tuned by R software; **: The parameters ranges were automatically selected by R software