Table 1 Classification performance of the self-similar behavior-based method (with 1024 window size, and six features and wavelet decomposition levels) and feature engineering method (22 features defined in Das and Khanna4.
From: Wavelet-based approach for diagnosing attention deficit hyperactivity disorder (ADHD)
Classification | Classification model | |||||
|---|---|---|---|---|---|---|
LR | SVM | KNN | ||||
Method | Sensitivity | Specificity | Sensitivity | Specificity | Sensitivity | Specificity |
Self-similarity | 0.92 | 0.71 | 0.97 | 0.72 | 0.95 | 0.65 |
Feature engineering | 0.65 | 0.73 | 0.63 | 0.77 | 0.66 | 0.66 |
Accuracy | Accuracy | Accuracy | ||||
Self-similarity | 78.56 | 84.30 | 80.15 | |||
Feature engineering | 67.54 | 69.58 | 66.43 | |||
