evaluation/feature_selection_justification.md

Feature selection justification

The face_orientation model uses 10 of 17 extracted features. This document summarises empirical support.

1. Domain rationale

The 10 features were chosen to cover three channels:

• Head pose: head_deviation, s_face, pitch
• Eye state: ear_left, ear_right, ear_avg, perclos
• Gaze: h_gaze, gaze_offset, s_eye

Excluded: v_gaze (noisy), mar (rare events), yaw/roll (redundant with head_deviation/s_face), blink_rate/closure_duration/yawn_duration (temporal overlap with perclos).

2. XGBoost feature importance (gain)

Config used: {'n_estimators': 600, 'max_depth': 8, 'learning_rate': 0.1489, 'subsample': 0.9625, 'colsample_bytree': 0.9013, 'reg_alpha': 1.1407, 'reg_lambda': 2.4181, 'eval_metric': 'logloss'}. Quick mode: yes (200 trees)

From the trained XGBoost checkpoint (gain on the 10 features):

Feature	Gain
head_deviation	8.83
s_face	10.27
s_eye	2.18
h_gaze	4.99
pitch	4.64
ear_left	3.57
ear_avg	6.96
ear_right	9.54
gaze_offset	1.80
perclos	5.68

Top 5 by gain: s_face, ear_right, head_deviation, ear_avg, perclos.

3. Leave-one-feature-out ablation (LOPO)

Baseline (all 10 features) mean LOPO F1: 0.8286.

Skipped in this run (--skip-lofo).

4. Channel ablation (LOPO)

Subset	Mean LOPO F1
head_pose	0.7480
eye_state	0.8071
gaze	0.7260
all_10	0.8286

5. Conclusion

Selection is supported by (1) domain rationale (three attention channels), (2) XGBoost gain importance, and (3) channel ablation. Run without --skip-lofo for full leave-one-out ablation.