Table 6 This table represents the classification models based on SNA, SNB, and ML-NSL angles for the diagnosis of skeletal class I/ III.

From: Machine learning models for improving the diagnosing efficiency of skeletal class I and III in German orthodontic patients

 

SNA

SNB

ML-NSL

Hyperparameters

Mean accuracy

Mean kappa

Balanced models hyperparameters

Balanced model (s) mean accuracy

Balanced model (s) mean Kappa

LDA

(+)

( +)

( +)

 

0.92

0.81

 

0.95

0.91

CART

(+)

(+)

(+)

 

0.759

0.41

 

0.68

0.36

KNN

(+)

(+)

(+)

KNN (k = 9)

0.88

0.64

KNN (k = 7)

0.91

0.83

SVM

(+)

(+)

(+)

SVM (Kernel type = Radial Basis Function, Sigma = 0.65, C = 1)

0.92

0.81

SVM (Kernel type = Radial Basis Function, Sigma = 0.81, C = 1)

0.92

0.85

RF

(+)

(+)

(+)

 

0.88

0.75

 

0.85

0.71

GLM

(+)

(+)

(+)

 

0.988

0.97

 

0.99

0.98

  1. For each model (LDA, CART, KNN, SVM, RF, GLM), the table summarized the accuracy and reliability (kappa) of six different machine learning models in diagnosing skeletal class I/III on the original sample size and on the down sampled data. The results in this table are obtained from the cross-validation data.
Back to article page