@@ -36,24 +36,32 @@ bootstrap_delta:
3636 $(PYTHON ) -m src.evaluation.bootstrap_eval \
3737 --model-a artifacts/models/multimodal_pneumonia_densenet121_triage_u_ignore_temporal_stronger_lr_v3/test_predictions.csv \
3838 --model-b artifacts/models/image_pneumonia_finetune_densenet121_u_ignore_temporal_stronger_lr_v3/test_predictions.csv \
39- --output-json artifacts/evaluation/bootstrap_multimodal_vs_image .json \
39+ --output-json artifacts/evaluation/bootstrap_multimodal_vs_image_stronger_lr_v3 .json \
4040 --n-bootstrap 2000 --seed $(SEED )
4141
4242calibration :
4343 $(PYTHON ) -m src.evaluation.calibration_analysis \
44- --output-dir artifacts/evaluation/calibration_final \
44+ --output-dir artifacts/evaluation/calibration_stronger_lr_v3 \
4545 --n-bins 10 --bootstrap --n-bootstrap 2000 \
4646 --model " Image" artifacts/models/image_pneumonia_finetune_densenet121_u_ignore_temporal_stronger_lr_v3/test_predictions.csv \
4747 --model " Multimodal" artifacts/models/multimodal_pneumonia_densenet121_triage_u_ignore_temporal_stronger_lr_v3/test_predictions.csv
4848
49+ dca :
50+ $(PYTHON ) -m src.evaluation.decision_curve_analysis \
51+ --output-dir artifacts/evaluation/dca \
52+ --model " Image" artifacts/models/image_pneumonia_finetune_densenet121_u_ignore_temporal_stronger_lr_v3/test_predictions.csv \
53+ --model " Multimodal" artifacts/models/multimodal_pneumonia_densenet121_triage_u_ignore_temporal_stronger_lr_v3/test_predictions.csv
54+
4955feature_ablation :
5056 $(PYTHON ) scripts/collect_feature_ablation_results.py
5157
52- evaluate : bootstrap_delta calibration feature_ablation
58+ evaluate : bootstrap_delta calibration dca feature_ablation
5359
5460# ─── SHAP ────────────────────────────────────────────────────────────────────
5561shap :
56- $(PYTHON ) scripts/generate_shap_clinical.py
62+ $(PYTHON ) scripts/generate_shap_clinical.py \
63+ --model-dir artifacts/models/clinical_xgb_u_ignore_temporal_strong_v2 \
64+ --feature-groups all
5765
5866# ─── Publication report ───────────────────────────────────────────────────────
5967report :
@@ -68,4 +76,4 @@ all: pretrain finetune_image finetune_multimodal train_clinical evaluate shap re
6876
6977.PHONY : preprocess preprocess_labs pretrain finetune_image finetune_multimodal \
7078 train_clinical_lr train_clinical_xgb train_clinical bootstrap_delta \
71- calibration feature_ablation evaluate shap report all test
79+ calibration dca feature_ablation evaluate shap report all test
0 commit comments