Table 4 Average classification accuracy for the different combinations of classifiers and types of features among all feature set sizes.

From: A data mining approach using cortical thickness for diagnosis and characterization of essential tremor

Classifier

All

Thickness

Roughness

Volume

Naive Bayes

0.6356 a  ± 0.0664

0.6547 a,b  ± 0.0285

0.6819 b  ± 0.0653

0.5627 c  ± 0.0388

Support Vector Machine

0.4570 a  ± 0.0823

0.6231 b  ± .0.0485

0.5648 c  ± 0.1208

0.4537 a  ± 0.0899

Rule

0.4532 a  ± 0.0661

0.5806 b  ± 0.0452

0.7092 c  ± 0.0661

0.5948 b  ± 0.0665

K-Nearest Neighbor

0.4150 a  ± 0.0688

0.5959 b  ± 0.0634

0.5773 b  ± 0.0476

0.4156 a  ± 0.0687

Artificial Neural Network

0.6067 a  ± 0.0922

0.5294 b  ± 0.0486

0.5534 b  ± 0.0979

0.5365 b  ± 0.0878

Average

0.5135 a  ± 0.1002

0.5967 a  ± 0.0470

0.6173 a  ± 0.0726

0.5127 a  ± 0.0754

  1. Accuracy range is between 0 and 1. Numbers with no shared subindex (a, b, c) presented statistically significant difference (p < 0.05, Student’s t-test, Bonferroni-corrected).
Back to article page