Skip to main content

Table 4 Settings and parameters for classification methods

From: Analysis of disease comorbidity patterns in a large-scale China population

LRusing L2 regularization norm
regularization intensity = 1
SVMusing the linear kernel function
penalty parameter of the error term = 10
RFDecision tree = 180
Bootstrap Sample
oob_score = true
Feature = Gini coefficient
NNUsing multilayer feedforward neural network
learning rate = 0.001
maximum number of iterations = 200
two hidden layers
randomly optimizing the size of mini batches