Table 8 Performance metrics of the best performing classifier (‘fair’ case) when the gender is used as a criterion to select the subjects of the training subset.
Dataset | Features and algorithm | Subjects included in the training subset | Se (%) | Sp (%) | \(\sqrt{Se\cdot Sp}\) (%) |
|---|---|---|---|---|---|
DOFDA | HCTSA features Naive Bayes (Gaussian) | Male subjects (testing with females) | 100.00 | 100.00 | 100.00 |
Random selection of users | 97.38 | 100.00 | 98.67 | ||
All (fair distribution) | 97.37 | 100.00 | 98.67 | ||
Female subjects (testing with males) | 95.26 | 100.00 | 97.60 | ||
Erciyes | Own selection of features SVM (quadratic kernel) | All (fair distribution) | 99.62 | 99.18 | 99.40 |
Random selection of users | 97.83 | 98.43 | 98.12 | ||
Female subjects (testing with males) | 97.29 | 98.06 | 97.68 | ||
Male subjects (testing with females) | 96.53 | 98.15 | 97.34 | ||
SisFall | HCTSA features SVM (cubic kernel) | Male subjects (testing with females) | 100.00 | 99.92 | 99.96 |
All (fair distribution) | 99.78 | 99.96 | 99.87 | ||
Random selection of users | 99.74 | 99.96 | 99.85 | ||
Female subjects (testing with males) | 99.00 | 99.85 | 99.42 | ||
UMAFall | Own selection of features KNN (Euclidean. 10 neighbors) | All (fair distribution) | 98.93 | 98.73 | 98.83 |
Random selection of users | 98.28 | 97.05 | 97.66 | ||
Male subjects (testing with females) | 95.35 | 98.02 | 96.68 | ||
Female subjects (testing with males) | 97.93 | 91.81 | 94.82 | ||
UP-Fall | Own selection of features SVM (linear kernel) | All (fair distribution) | 99.59 | 98.02 | 98.80 |
Male subjects (testing with females) | 99.10 | 98.31 | 98.70 | ||
Random selection of users | 99.65 | 97.56 | 98.60 | ||
Female subjects (testing with males) | 98.51 | 95.56 | 97.02 |
