A Novel Approach for Feature Selection and Classification of Diabetes Mellitus: Machine Learning Methods

Table 1

Hyperparameter optimization.

K-nearest neighbour	Random forest	Decision trees	Multilayer perceptron
Number of neighbours = 45	Size of each bag = 53	Confidence factor = 0.11	Learning rate = 0.003
Batch size = 100	Max depth = 0	Min num. of objects = 1	Momentum = 0.9
Algorithm = linear search	No. of trees = 100	Unpruned = false	Hidden layers = 10
Distance function = Manhattan function