Table 4.
Performance of ML with OSCAR, SLOF as targets for SZ group. Gini was used to rank top features. Input variables considered were 120 ER40(cognitive: 40 RT, 40 corr and meta-cognitive: 40 CR). Target variable considered were categorical (0,1) while all others were numeric. Here threshold refers to the value of target used for categorization. Performance is shown for the best ranked features and model.
| Target | Threshold | Method | Number of ranked featuresa, b | Count of (CR)a, b features | Count of negative featuresa | Count of negative (CR)a features | F1 | AUC | Equivalent Cohen's dc |
|---|---|---|---|---|---|---|---|---|---|
| Oscars Self Report | 0–17 as 0, 18 and above as 1 | Stochastic Gradient Descent (SGD) | 80 | 34 | 49 | 22 | 0.74 | 0.74 | 0.91 |
| Oscars Informant Report | 0–17 as 0, 18 and above as 1 | AdaBoost | 40 | 27 | 29 | 18 | 0.81 | 0.73 | 0.87 |
a
Negative features considered for ER40 are S, A and F.
b
Positive features considered for ER40 are N, H.
c
Table 2 from (Salgado, 2018).