psqi_lcga_group_2
CPT_REACTION_TIME_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'CPT_REACTION_TIME_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
NaiveBayes 0.717 0.462 0.556 0.395 0.850 0.811 0.892 0.324 0.765 5.400 24.400
RandomForest 0.692 0.441 0.619 0.342 0.862 0.805 0.928 0.338 0.779 4.200 25.600
LogisticRegression 0.690 0.512 0.458 0.579 0.802 0.842 0.766 0.322 0.718 9.600 20.200
KNN 0.686 0.433 0.591 0.342 0.857 0.803 0.919 0.321 0.772 4.400 25.400
SVM 0.682 0.530 0.489 0.579 0.819 0.846 0.793 0.353 0.738 9.000 20.800
MLP 0.657 0.416 0.410 0.421 0.796 0.800 0.793 0.212 0.698 7.800 22.000
XGBoost 0.651 0.472 0.500 0.447 0.832 0.817 0.847 0.306 0.745 6.800 23.000
DecisionTree 0.638 0.462 0.450 0.474 0.809 0.817 0.802 0.271 0.718 8.000 21.800
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
NaiveBayes 99 12 23 15
RandomForest 103 8 25 13
LogisticRegression 85 26 16 22
KNN 102 9 25 13
SVM 88 23 16 22
MLP 88 23 22 16
XGBoost 94 17 21 17
DecisionTree 89 22 20 18
CPT_DPR_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'CPT_DPR_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.709 0.512 0.477 0.553 0.815 0.838 0.793 0.330 0.732 8.800 21.000
NaiveBayes 0.702 0.492 0.593 0.421 0.858 0.820 0.901 0.364 0.779 5.400 24.400
SVM 0.670 0.494 0.447 0.553 0.798 0.833 0.766 0.299 0.711 9.400 20.400
RandomForest 0.622 0.492 0.652 0.395 0.869 0.817 0.928 0.389 0.792 4.600 25.200
KNN 0.618 0.431 0.519 0.368 0.841 0.803 0.883 0.284 0.752 5.400 24.400
DecisionTree 0.595 0.400 0.381 0.421 0.780 0.794 0.766 0.181 0.678 8.400 21.400
XGBoost 0.577 0.366 0.394 0.342 0.802 0.784 0.820 0.170 0.698 6.600 23.200
MLP 0.556 0.421 0.421 0.421 0.802 0.802 0.802 0.223 0.705 7.600 22.200
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 88 23 17 21
NaiveBayes 100 11 22 16
SVM 85 26 17 21
RandomForest 103 8 23 15
KNN 98 13 24 14
DecisionTree 85 26 22 16
XGBoost 91 20 25 13
MLP 89 22 22 16
CPT_OMISSION_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'CPT_OMISSION_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
SVM 0.705 0.539 0.471 0.632 0.804 0.857 0.757 0.357 0.725 10.200 19.600
LogisticRegression 0.691 0.541 0.489 0.605 0.817 0.853 0.784 0.365 0.738 9.400 20.400
KNN 0.691 0.452 0.583 0.368 0.856 0.808 0.910 0.330 0.772 4.800 25.000
NaiveBayes 0.657 0.478 0.552 0.421 0.848 0.817 0.883 0.335 0.765 5.800 24.000
MLP 0.648 0.548 0.571 0.526 0.853 0.842 0.865 0.402 0.779 7.000 22.800
RandomForest 0.646 0.424 0.500 0.368 0.836 0.802 0.874 0.270 0.745 5.600 24.200
XGBoost 0.614 0.450 0.429 0.474 0.798 0.813 0.784 0.249 0.705 8.400 21.400
DecisionTree 0.611 0.421 0.421 0.421 0.802 0.802 0.802 0.223 0.705 7.600 22.200
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
SVM 84 27 14 24
LogisticRegression 87 24 15 23
KNN 101 10 24 14
NaiveBayes 98 13 22 16
MLP 96 15 18 20
RandomForest 97 14 24 14
XGBoost 87 24 20 18
DecisionTree 89 22 22 16
CPT_COMMISSION_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'CPT_COMMISSION_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.715 0.482 0.444 0.526 0.800 0.827 0.775 0.286 0.711 9.000 20.800
KNN 0.709 0.364 0.588 0.263 0.856 0.788 0.937 0.274 0.765 3.400 26.400
RandomForest 0.704 0.431 0.519 0.368 0.841 0.803 0.883 0.284 0.752 5.400 24.400
NaiveBayes 0.701 0.469 0.577 0.395 0.855 0.813 0.901 0.340 0.772 5.200 24.600
SVM 0.694 0.518 0.468 0.579 0.808 0.843 0.775 0.332 0.725 9.400 20.400
XGBoost 0.688 0.432 0.444 0.421 0.812 0.805 0.820 0.245 0.718 7.200 22.600
MLP 0.685 0.463 0.432 0.500 0.796 0.819 0.775 0.263 0.705 8.800 21.000
DecisionTree 0.631 0.456 0.439 0.474 0.804 0.815 0.793 0.260 0.711 8.200 21.600
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 86 25 18 20
KNN 104 7 28 10
RandomForest 98 13 24 14
NaiveBayes 100 11 23 15
SVM 86 25 16 22
XGBoost 91 20 22 16
MLP 86 25 19 19
DecisionTree 88 23 20 18
CPT_PRS_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'CPT_PRS_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
SVM 0.685 0.494 0.447 0.553 0.798 0.833 0.766 0.299 0.711 9.400 20.400
LogisticRegression 0.675 0.494 0.447 0.553 0.798 0.833 0.766 0.299 0.711 9.400 20.400
NaiveBayes 0.652 0.431 0.519 0.368 0.841 0.803 0.883 0.284 0.752 5.400 24.400
KNN 0.652 0.413 0.520 0.342 0.843 0.798 0.892 0.273 0.752 5.000 24.800
RandomForest 0.643 0.462 0.556 0.395 0.850 0.811 0.892 0.324 0.765 5.400 24.400
MLP 0.620 0.421 0.421 0.421 0.802 0.802 0.802 0.223 0.705 7.600 22.200
XGBoost 0.618 0.486 0.500 0.474 0.830 0.823 0.838 0.317 0.745 7.200 22.600
DecisionTree 0.608 0.439 0.409 0.474 0.787 0.810 0.766 0.229 0.691 8.800 21.000
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
SVM 85 26 17 21
LogisticRegression 85 26 17 21
NaiveBayes 98 13 24 14
KNN 99 12 25 13
RandomForest 99 12 23 15
MLP 89 22 22 16
XGBoost 93 18 20 18
DecisionTree 85 26 20 18
CPT_HRT_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'CPT_HRT_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
NaiveBayes 0.714 0.438 0.538 0.368 0.846 0.805 0.892 0.299 0.758 5.200 24.600
LogisticRegression 0.713 0.512 0.458 0.579 0.802 0.842 0.766 0.322 0.718 9.600 20.200
SVM 0.671 0.506 0.449 0.579 0.796 0.840 0.757 0.311 0.711 9.800 20.000
RandomForest 0.659 0.444 0.560 0.368 0.851 0.806 0.901 0.314 0.765 5.000 24.800
XGBoost 0.651 0.390 0.385 0.395 0.787 0.791 0.784 0.177 0.685 7.800 22.000
DecisionTree 0.638 0.462 0.450 0.474 0.809 0.817 0.802 0.271 0.718 8.000 21.800
KNN 0.621 0.373 0.524 0.289 0.845 0.789 0.910 0.250 0.752 4.200 25.600
MLP 0.573 0.361 0.382 0.342 0.796 0.783 0.811 0.159 0.691 6.800 23.000
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
NaiveBayes 99 12 24 14
LogisticRegression 85 26 16 22
SVM 84 27 16 22
RandomForest 100 11 24 14
XGBoost 87 24 23 15
DecisionTree 89 22 20 18
KNN 101 10 27 11
MLP 90 21 25 13
CPT_HRTSD_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'CPT_HRTSD_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
SVM 0.726 0.529 0.469 0.605 0.806 0.850 0.766 0.344 0.725 9.800 20.000
MLP 0.713 0.475 0.452 0.500 0.807 0.822 0.793 0.284 0.718 8.400 21.400
LogisticRegression 0.709 0.541 0.489 0.605 0.817 0.853 0.784 0.365 0.738 9.400 20.400
NaiveBayes 0.694 0.444 0.560 0.368 0.851 0.806 0.901 0.314 0.765 5.000 24.800
KNN 0.689 0.393 0.522 0.316 0.844 0.794 0.901 0.261 0.752 4.600 25.200
RandomForest 0.677 0.484 0.625 0.395 0.864 0.816 0.919 0.372 0.785 4.800 25.000
XGBoost 0.658 0.456 0.439 0.474 0.804 0.815 0.793 0.260 0.711 8.200 21.600
DecisionTree 0.651 0.481 0.463 0.500 0.813 0.824 0.802 0.295 0.725 8.200 21.600
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
SVM 85 26 15 23
MLP 88 23 19 19
LogisticRegression 87 24 15 23
NaiveBayes 100 11 24 14
KNN 100 11 26 12
RandomForest 102 9 23 15
XGBoost 88 23 20 18
DecisionTree 89 22 19 19
CPT_VAR_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'CPT_VAR_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.697 0.516 0.436 0.632 0.780 0.851 0.721 0.318 0.698 11.000 18.800
NaiveBayes 0.687 0.485 0.571 0.421 0.853 0.818 0.892 0.349 0.772 5.600 24.200
SVM 0.684 0.512 0.458 0.579 0.802 0.842 0.766 0.322 0.718 9.600 20.200
RandomForest 0.678 0.444 0.560 0.368 0.851 0.806 0.901 0.314 0.765 5.000 24.800
KNN 0.658 0.375 0.462 0.316 0.829 0.789 0.874 0.218 0.732 5.200 24.600
XGBoost 0.647 0.459 0.472 0.447 0.821 0.814 0.829 0.281 0.732 7.200 22.600
DecisionTree 0.609 0.420 0.395 0.447 0.783 0.802 0.766 0.205 0.685 8.600 21.200
MLP 0.551 0.321 0.302 0.342 0.747 0.764 0.730 0.069 0.631 8.600 21.200
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 80 31 14 24
NaiveBayes 99 12 22 16
SVM 85 26 16 22
RandomForest 100 11 24 14
KNN 97 14 26 12
XGBoost 92 19 21 17
DecisionTree 85 26 21 17
MLP 81 30 25 13
CPT_BLOCK_CHANGE_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'CPT_BLOCK_CHANGE_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.715 0.494 0.447 0.553 0.798 0.833 0.766 0.299 0.711 9.400 20.400
NaiveBayes 0.689 0.444 0.560 0.368 0.851 0.806 0.901 0.314 0.765 5.000 24.800
KNN 0.640 0.321 0.500 0.237 0.843 0.779 0.919 0.208 0.745 3.600 26.200
RandomForest 0.639 0.424 0.500 0.368 0.836 0.802 0.874 0.270 0.745 5.600 24.200
SVM 0.620 0.468 0.393 0.579 0.755 0.828 0.694 0.245 0.664 11.200 18.600
DecisionTree 0.603 0.400 0.381 0.421 0.780 0.794 0.766 0.181 0.678 8.400 21.400
MLP 0.595 0.366 0.394 0.342 0.802 0.784 0.820 0.170 0.698 6.600 23.200
XGBoost 0.566 0.375 0.357 0.395 0.771 0.785 0.757 0.147 0.664 8.400 21.400
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 85 26 17 21
NaiveBayes 100 11 24 14
KNN 102 9 29 9
RandomForest 97 14 24 14
SVM 77 34 16 22
DecisionTree 85 26 22 16
MLP 91 20 25 13
XGBoost 84 27 23 15
CPT_ISI_CHANGE_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'CPT_ISI_CHANGE_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
NaiveBayes 0.726 0.438 0.538 0.368 0.846 0.805 0.892 0.299 0.758 5.200 24.600
RandomForest 0.711 0.355 0.458 0.289 0.831 0.784 0.883 0.204 0.732 4.800 25.000
LogisticRegression 0.708 0.524 0.478 0.579 0.813 0.845 0.784 0.342 0.732 9.200 20.600
SVM 0.688 0.516 0.436 0.632 0.780 0.851 0.721 0.318 0.698 11.000 18.800
XGBoost 0.633 0.400 0.405 0.395 0.798 0.795 0.802 0.198 0.698 7.400 22.400
DecisionTree 0.625 0.442 0.436 0.447 0.805 0.809 0.802 0.247 0.711 7.800 22.000
KNN 0.601 0.393 0.522 0.316 0.844 0.794 0.901 0.261 0.752 4.600 25.200
MLP 0.534 0.314 0.344 0.289 0.789 0.769 0.811 0.106 0.678 6.400 23.400
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
NaiveBayes 99 12 24 14
RandomForest 98 13 27 11
LogisticRegression 87 24 16 22
SVM 80 31 14 24
XGBoost 89 22 23 15
DecisionTree 89 22 21 17
KNN 100 11 26 12
MLP 90 21 27 11
IGT_NET_TOTAL
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_NET_TOTAL']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.701 0.535 0.479 0.605 0.811 0.851 0.775 0.354 0.732 9.600 20.200
NaiveBayes 0.698 0.469 0.577 0.395 0.855 0.813 0.901 0.340 0.772 5.200 24.600
SVM 0.633 0.494 0.487 0.500 0.824 0.827 0.820 0.317 0.738 7.800 22.000
KNN 0.624 0.436 0.706 0.316 0.872 0.803 0.955 0.371 0.792 3.400 26.400
RandomForest 0.593 0.369 0.444 0.316 0.824 0.787 0.865 0.204 0.725 5.400 24.400
MLP 0.587 0.368 0.368 0.368 0.784 0.784 0.784 0.152 0.678 7.600 22.200
DecisionTree 0.557 0.346 0.326 0.368 0.756 0.774 0.739 0.103 0.644 8.600 21.200
XGBoost 0.540 0.378 0.389 0.368 0.795 0.788 0.802 0.173 0.691 7.200 22.600
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 86 25 15 23
NaiveBayes 100 11 23 15
SVM 91 20 19 19
KNN 106 5 26 12
RandomForest 96 15 26 12
MLP 87 24 24 14
DecisionTree 82 29 24 14
XGBoost 89 22 24 14
IGT_NET_1
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_NET_1']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
NaiveBayes 0.706 0.444 0.560 0.368 0.851 0.806 0.901 0.314 0.765 5.000 24.800
LogisticRegression 0.690 0.494 0.447 0.553 0.798 0.833 0.766 0.299 0.711 9.400 20.400
RandomForest 0.651 0.413 0.520 0.342 0.843 0.798 0.892 0.273 0.752 5.000 24.800
SVM 0.643 0.500 0.457 0.553 0.804 0.835 0.775 0.309 0.718 9.200 20.600
KNN 0.630 0.373 0.524 0.289 0.845 0.789 0.910 0.250 0.752 4.200 25.600
DecisionTree 0.620 0.439 0.409 0.474 0.787 0.810 0.766 0.229 0.691 8.800 21.000
XGBoost 0.570 0.405 0.390 0.421 0.785 0.796 0.775 0.191 0.685 8.200 21.600
MLP 0.541 0.400 0.405 0.395 0.798 0.795 0.802 0.198 0.698 7.400 22.400
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
NaiveBayes 100 11 24 14
LogisticRegression 85 26 17 21
RandomForest 99 12 25 13
SVM 86 25 17 21
KNN 101 10 27 11
DecisionTree 85 26 20 18
XGBoost 86 25 22 16
MLP 89 22 23 15
IGT_NET_2
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_NET_2']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.710 0.512 0.458 0.579 0.802 0.842 0.766 0.322 0.718 9.600 20.200
SVM 0.699 0.500 0.457 0.553 0.804 0.835 0.775 0.309 0.718 9.200 20.600
NaiveBayes 0.688 0.469 0.577 0.395 0.855 0.813 0.901 0.340 0.772 5.200 24.600
KNN 0.667 0.373 0.524 0.289 0.845 0.789 0.910 0.250 0.752 4.200 25.600
RandomForest 0.655 0.393 0.522 0.316 0.844 0.794 0.901 0.261 0.752 4.600 25.200
DecisionTree 0.622 0.427 0.432 0.421 0.807 0.804 0.811 0.234 0.711 7.400 22.400
XGBoost 0.611 0.364 0.359 0.368 0.778 0.782 0.775 0.142 0.671 7.800 22.000
MLP 0.606 0.361 0.382 0.342 0.796 0.783 0.811 0.159 0.691 6.800 23.000
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 85 26 16 22
SVM 86 25 17 21
NaiveBayes 100 11 23 15
KNN 101 10 27 11
RandomForest 100 11 26 12
DecisionTree 90 21 22 16
XGBoost 86 25 24 14
MLP 90 21 25 13
IGT_NET_3
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_NET_3']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
NaiveBayes 0.714 0.462 0.556 0.395 0.850 0.811 0.892 0.324 0.765 5.400 24.400
LogisticRegression 0.709 0.539 0.471 0.632 0.804 0.857 0.757 0.357 0.725 10.200 19.600
DecisionTree 0.647 0.474 0.474 0.474 0.820 0.820 0.820 0.294 0.732 7.600 22.200
RandomForest 0.634 0.418 0.483 0.368 0.831 0.800 0.865 0.257 0.738 5.800 24.000
SVM 0.629 0.475 0.452 0.500 0.807 0.822 0.793 0.284 0.718 8.400 21.400
XGBoost 0.624 0.453 0.459 0.447 0.816 0.812 0.820 0.270 0.725 7.400 22.400
KNN 0.613 0.373 0.524 0.289 0.845 0.789 0.910 0.250 0.752 4.200 25.600
MLP 0.583 0.429 0.469 0.395 0.825 0.803 0.847 0.256 0.732 6.400 23.400
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
NaiveBayes 99 12 23 15
LogisticRegression 84 27 14 24
DecisionTree 91 20 20 18
RandomForest 96 15 24 14
SVM 88 23 19 19
XGBoost 91 20 21 17
KNN 101 10 27 11
MLP 94 17 23 15
IGT_NET_4
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_NET_4']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
NaiveBayes 0.722 0.484 0.625 0.395 0.864 0.816 0.919 0.372 0.785 4.800 25.000
LogisticRegression 0.713 0.518 0.468 0.579 0.808 0.843 0.775 0.332 0.725 9.400 20.400
SVM 0.675 0.494 0.447 0.553 0.798 0.833 0.766 0.299 0.711 9.400 20.400
DecisionTree 0.615 0.434 0.400 0.474 0.781 0.808 0.757 0.219 0.685 9.000 20.800
RandomForest 0.613 0.438 0.538 0.368 0.846 0.805 0.892 0.299 0.758 5.200 24.600
KNN 0.612 0.508 0.714 0.395 0.879 0.820 0.946 0.427 0.805 4.200 25.600
XGBoost 0.579 0.439 0.409 0.474 0.787 0.810 0.766 0.229 0.691 8.800 21.000
MLP 0.502 0.329 0.343 0.316 0.782 0.772 0.793 0.112 0.671 7.000 22.800
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
NaiveBayes 102 9 23 15
LogisticRegression 86 25 16 22
SVM 85 26 17 21
DecisionTree 84 27 20 18
RandomForest 99 12 24 14
KNN 105 6 23 15
XGBoost 85 26 20 18
MLP 88 23 26 12
IGT_NET_5
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_NET_5']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.709 0.500 0.440 0.579 0.790 0.838 0.748 0.302 0.705 10.000 19.800
NaiveBayes 0.694 0.444 0.560 0.368 0.851 0.806 0.901 0.314 0.765 5.000 24.800
KNN 0.676 0.310 0.450 0.237 0.833 0.775 0.901 0.176 0.732 4.000 25.800
RandomForest 0.641 0.381 0.480 0.316 0.834 0.790 0.883 0.232 0.738 5.000 24.800
SVM 0.624 0.479 0.397 0.605 0.752 0.835 0.685 0.259 0.664 11.600 18.200
MLP 0.600 0.361 0.382 0.342 0.796 0.783 0.811 0.159 0.691 6.800 23.000
XGBoost 0.595 0.405 0.390 0.421 0.785 0.796 0.775 0.191 0.685 8.200 21.600
DecisionTree 0.594 0.389 0.412 0.368 0.805 0.791 0.820 0.196 0.705 6.800 23.000
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 83 28 16 22
NaiveBayes 100 11 24 14
KNN 100 11 29 9
RandomForest 98 13 26 12
SVM 76 35 15 23
MLP 90 21 25 13
XGBoost 86 25 22 16
DecisionTree 91 20 24 14
IGT_NET_5_MINUS_1
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_NET_5_MINUS_1']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.709 0.506 0.449 0.579 0.796 0.840 0.757 0.311 0.711 9.800 20.000
MLP 0.685 0.487 0.475 0.500 0.818 0.826 0.811 0.306 0.732 8.000 21.800
NaiveBayes 0.680 0.469 0.577 0.395 0.855 0.813 0.901 0.340 0.772 5.200 24.600
SVM 0.667 0.522 0.444 0.632 0.786 0.853 0.730 0.328 0.705 10.800 19.000
KNN 0.664 0.381 0.480 0.316 0.834 0.790 0.883 0.232 0.738 5.000 24.800
RandomForest 0.652 0.455 0.536 0.395 0.845 0.810 0.883 0.310 0.758 5.600 24.200
DecisionTree 0.625 0.442 0.436 0.447 0.805 0.809 0.802 0.247 0.711 7.800 22.000
XGBoost 0.607 0.427 0.432 0.421 0.807 0.804 0.811 0.234 0.711 7.400 22.400
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 84 27 16 22
MLP 90 21 19 19
NaiveBayes 100 11 23 15
SVM 81 30 14 24
KNN 98 13 26 12
RandomForest 98 13 23 15
DecisionTree 89 22 21 17
XGBoost 90 21 22 16
IGT_DECK_A
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_DECK_A']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
NaiveBayes 0.690 0.444 0.560 0.368 0.851 0.806 0.901 0.314 0.765 5.000 24.800
LogisticRegression 0.687 0.523 0.460 0.605 0.800 0.848 0.757 0.334 0.718 10.000 19.800
RandomForest 0.618 0.364 0.429 0.316 0.819 0.785 0.856 0.192 0.718 5.600 24.200
XGBoost 0.598 0.406 0.452 0.368 0.821 0.797 0.847 0.231 0.725 6.200 23.600
SVM 0.596 0.472 0.412 0.553 0.775 0.827 0.730 0.259 0.685 10.200 19.600
DecisionTree 0.576 0.375 0.357 0.395 0.771 0.785 0.757 0.147 0.664 8.400 21.400
KNN 0.551 0.338 0.407 0.289 0.815 0.779 0.856 0.164 0.711 5.400 24.400
MLP 0.525 0.329 0.343 0.316 0.782 0.772 0.793 0.112 0.671 7.000 22.800
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
NaiveBayes 100 11 24 14
LogisticRegression 84 27 15 23
RandomForest 95 16 26 12
XGBoost 94 17 24 14
SVM 81 30 17 21
DecisionTree 84 27 23 15
KNN 95 16 27 11
MLP 88 23 26 12
IGT_DECK_B
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_DECK_B']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.707 0.518 0.468 0.579 0.808 0.843 0.775 0.332 0.725 9.400 20.400
NaiveBayes 0.691 0.444 0.560 0.368 0.851 0.806 0.901 0.314 0.765 5.000 24.800
SVM 0.662 0.512 0.477 0.553 0.815 0.838 0.793 0.330 0.732 8.800 21.000
RandomForest 0.609 0.441 0.500 0.395 0.835 0.807 0.865 0.282 0.745 6.000 23.800
KNN 0.597 0.308 0.571 0.211 0.854 0.778 0.946 0.234 0.758 2.800 27.000
XGBoost 0.560 0.439 0.409 0.474 0.787 0.810 0.766 0.229 0.691 8.800 21.000
MLP 0.545 0.400 0.438 0.368 0.816 0.795 0.838 0.219 0.718 6.400 23.400
DecisionTree 0.542 0.316 0.316 0.316 0.766 0.766 0.766 0.082 0.651 7.600 22.200
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 86 25 16 22
NaiveBayes 100 11 24 14
SVM 88 23 17 21
RandomForest 96 15 23 15
KNN 105 6 30 8
XGBoost 85 26 20 18
MLP 93 18 24 14
DecisionTree 85 26 26 12
IGT_DECK_C
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_DECK_C']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.703 0.506 0.449 0.579 0.796 0.840 0.757 0.311 0.711 9.800 20.000
NaiveBayes 0.693 0.444 0.560 0.368 0.851 0.806 0.901 0.314 0.765 5.000 24.800
KNN 0.685 0.406 0.500 0.342 0.838 0.797 0.883 0.258 0.745 5.200 24.600
MLP 0.644 0.411 0.429 0.395 0.809 0.798 0.820 0.221 0.711 7.000 22.800
RandomForest 0.618 0.381 0.480 0.316 0.834 0.790 0.883 0.232 0.738 5.000 24.800
SVM 0.617 0.468 0.393 0.579 0.755 0.828 0.694 0.245 0.664 11.200 18.600
XGBoost 0.582 0.415 0.386 0.447 0.778 0.800 0.757 0.195 0.678 8.800 21.000
DecisionTree 0.547 0.320 0.324 0.316 0.771 0.768 0.775 0.091 0.658 7.400 22.400
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 84 27 16 22
NaiveBayes 100 11 24 14
KNN 98 13 25 13
MLP 91 20 23 15
RandomForest 98 13 26 12
SVM 77 34 16 22
XGBoost 84 27 21 17
DecisionTree 86 25 26 12
IGT_DECK_D
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'IGT_DECK_D']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
KNN 0.722 0.407 0.571 0.316 0.854 0.797 0.919 0.294 0.765 4.200 25.600
LogisticRegression 0.712 0.529 0.469 0.605 0.806 0.850 0.766 0.344 0.725 9.800 20.000
MLP 0.711 0.538 0.525 0.553 0.836 0.844 0.829 0.375 0.758 8.000 21.800
NaiveBayes 0.694 0.444 0.560 0.368 0.851 0.806 0.901 0.314 0.765 5.000 24.800
RandomForest 0.688 0.431 0.519 0.368 0.841 0.803 0.883 0.284 0.752 5.400 24.400
SVM 0.687 0.558 0.500 0.632 0.821 0.861 0.784 0.387 0.745 9.600 20.200
DecisionTree 0.655 0.488 0.455 0.526 0.806 0.829 0.784 0.296 0.718 8.800 21.000
XGBoost 0.599 0.474 0.474 0.474 0.820 0.820 0.820 0.294 0.732 7.600 22.200
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
KNN 102 9 26 12
LogisticRegression 85 26 15 23
MLP 92 19 17 21
NaiveBayes 100 11 24 14
RandomForest 98 13 24 14
SVM 87 24 14 24
DecisionTree 87 24 18 20
XGBoost 91 20 20 18
WCST_TOTAL_ERRORS_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'WCST_TOTAL_ERRORS_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.714 0.562 0.490 0.658 0.813 0.867 0.766 0.389 0.738 10.200 19.600
NaiveBayes 0.694 0.418 0.483 0.368 0.831 0.800 0.865 0.257 0.738 5.800 24.000
SVM 0.693 0.525 0.426 0.684 0.764 0.864 0.685 0.327 0.685 12.200 17.600
RandomForest 0.669 0.406 0.500 0.342 0.838 0.797 0.883 0.258 0.745 5.200 24.600
DecisionTree 0.651 0.482 0.444 0.526 0.800 0.827 0.775 0.286 0.711 9.000 20.800
KNN 0.649 0.233 0.318 0.184 0.807 0.756 0.865 0.060 0.691 4.400 25.400
XGBoost 0.578 0.385 0.375 0.395 0.782 0.789 0.775 0.167 0.678 8.000 21.800
MLP 0.565 0.427 0.432 0.421 0.807 0.804 0.811 0.234 0.711 7.400 22.400
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 85 26 13 25
NaiveBayes 96 15 24 14
SVM 76 35 12 26
RandomForest 98 13 25 13
DecisionTree 86 25 18 20
KNN 96 15 31 7
XGBoost 86 25 23 15
MLP 90 21 22 16
WCST_PCT_ERRORS_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'WCST_PCT_ERRORS_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.729 0.552 0.490 0.632 0.815 0.860 0.775 0.377 0.738 9.800 20.000
NaiveBayes 0.717 0.431 0.519 0.368 0.841 0.803 0.883 0.284 0.752 5.400 24.400
SVM 0.686 0.536 0.441 0.684 0.776 0.867 0.703 0.345 0.698 11.800 18.000
KNN 0.669 0.276 0.400 0.211 0.825 0.767 0.892 0.131 0.718 4.000 25.800
RandomForest 0.655 0.426 0.565 0.342 0.852 0.802 0.910 0.304 0.765 4.600 25.200
DecisionTree 0.616 0.430 0.415 0.447 0.795 0.806 0.784 0.226 0.698 8.200 21.600
MLP 0.603 0.400 0.438 0.368 0.816 0.795 0.838 0.219 0.718 6.400 23.400
XGBoost 0.553 0.375 0.357 0.395 0.771 0.785 0.757 0.147 0.664 8.400 21.400
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 86 25 14 24
NaiveBayes 98 13 24 14
SVM 78 33 12 26
KNN 99 12 30 8
RandomForest 101 10 25 13
DecisionTree 87 24 21 17
MLP 93 18 24 14
XGBoost 84 27 23 15
WCST_PERS_RESP_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'WCST_PERS_RESP_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.732 0.549 0.472 0.658 0.802 0.865 0.748 0.369 0.725 10.600 19.200
NaiveBayes 0.712 0.438 0.538 0.368 0.846 0.805 0.892 0.299 0.758 5.200 24.600
KNN 0.684 0.344 0.423 0.289 0.821 0.780 0.865 0.177 0.718 5.200 24.600
SVM 0.677 0.510 0.406 0.684 0.745 0.859 0.658 0.301 0.664 12.800 17.000
RandomForest 0.647 0.406 0.500 0.342 0.838 0.797 0.883 0.258 0.745 5.200 24.600
MLP 0.632 0.378 0.389 0.368 0.795 0.788 0.802 0.173 0.691 7.200 22.600
XGBoost 0.631 0.419 0.375 0.474 0.764 0.802 0.730 0.190 0.664 9.600 20.200
DecisionTree 0.563 0.347 0.351 0.342 0.780 0.777 0.784 0.127 0.671 7.400 22.400
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 83 28 13 25
NaiveBayes 99 12 24 14
KNN 96 15 27 11
SVM 73 38 12 26
RandomForest 98 13 25 13
MLP 89 22 24 14
XGBoost 81 30 20 18
DecisionTree 87 24 25 13
WCST_PCT_PERS_RESP_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'WCST_PCT_PERS_RESP_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.740 0.556 0.481 0.658 0.808 0.866 0.757 0.379 0.732 10.400 19.400
NaiveBayes 0.717 0.462 0.556 0.395 0.850 0.811 0.892 0.324 0.765 5.400 24.400
KNN 0.688 0.344 0.423 0.289 0.821 0.780 0.865 0.177 0.718 5.200 24.600
SVM 0.658 0.465 0.377 0.605 0.734 0.830 0.658 0.233 0.644 12.200 17.600
RandomForest 0.631 0.355 0.458 0.289 0.831 0.784 0.883 0.204 0.732 4.800 25.000
XGBoost 0.607 0.370 0.349 0.395 0.765 0.783 0.748 0.137 0.658 8.600 21.200
MLP 0.600 0.442 0.436 0.447 0.805 0.809 0.802 0.247 0.711 7.800 22.000
DecisionTree 0.589 0.390 0.385 0.395 0.787 0.791 0.784 0.177 0.685 7.800 22.000
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 84 27 13 25
NaiveBayes 99 12 23 15
KNN 96 15 27 11
SVM 73 38 15 23
RandomForest 98 13 27 11
XGBoost 83 28 23 15
MLP 89 22 21 17
DecisionTree 87 24 23 15
WCST_PERS_ERR_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'WCST_PERS_ERR_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.730 0.539 0.471 0.632 0.804 0.857 0.757 0.357 0.725 10.200 19.600
NaiveBayes 0.715 0.438 0.538 0.368 0.846 0.805 0.892 0.299 0.758 5.200 24.600
KNN 0.709 0.344 0.423 0.289 0.821 0.780 0.865 0.177 0.718 5.200 24.600
SVM 0.681 0.510 0.406 0.684 0.745 0.859 0.658 0.301 0.664 12.800 17.000
MLP 0.657 0.406 0.452 0.368 0.821 0.797 0.847 0.231 0.725 6.200 23.600
RandomForest 0.643 0.393 0.522 0.316 0.844 0.794 0.901 0.261 0.752 4.600 25.200
XGBoost 0.623 0.444 0.419 0.474 0.793 0.811 0.775 0.239 0.698 8.600 21.200
DecisionTree 0.569 0.351 0.361 0.342 0.786 0.779 0.793 0.137 0.678 7.200 22.600
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 84 27 14 24
NaiveBayes 99 12 24 14
KNN 96 15 27 11
SVM 73 38 12 26
MLP 94 17 24 14
RandomForest 100 11 26 12
XGBoost 86 25 20 18
DecisionTree 88 23 25 13
WCST_PCT_PERS_ERR_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'WCST_PCT_PERS_ERR_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.733 0.539 0.471 0.632 0.804 0.857 0.757 0.357 0.725 10.200 19.600
NaiveBayes 0.723 0.462 0.556 0.395 0.850 0.811 0.892 0.324 0.765 5.400 24.400
KNN 0.662 0.308 0.370 0.263 0.807 0.770 0.847 0.124 0.698 5.400 24.400
SVM 0.646 0.495 0.397 0.658 0.741 0.849 0.658 0.278 0.658 12.600 17.200
DecisionTree 0.634 0.451 0.485 0.421 0.828 0.810 0.847 0.281 0.738 6.600 23.200
RandomForest 0.631 0.400 0.481 0.342 0.833 0.795 0.874 0.244 0.738 5.400 24.400
MLP 0.597 0.456 0.439 0.474 0.804 0.815 0.793 0.260 0.711 8.200 21.600
XGBoost 0.589 0.395 0.372 0.421 0.774 0.792 0.757 0.171 0.671 8.600 21.200
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 84 27 14 24
NaiveBayes 99 12 23 15
KNN 94 17 28 10
SVM 73 38 13 25
DecisionTree 94 17 22 16
RandomForest 97 14 25 13
MLP 88 23 20 18
XGBoost 84 27 22 16
WCST_NONPERS_ERR_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'WCST_NONPERS_ERR_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
NaiveBayes 0.732 0.418 0.483 0.368 0.831 0.800 0.865 0.257 0.738 5.800 24.000
LogisticRegression 0.729 0.552 0.490 0.632 0.815 0.860 0.775 0.377 0.738 9.800 20.000
SVM 0.698 0.539 0.471 0.632 0.804 0.857 0.757 0.357 0.725 10.200 19.600
DecisionTree 0.676 0.506 0.488 0.526 0.822 0.833 0.811 0.329 0.738 8.200 21.600
RandomForest 0.649 0.438 0.538 0.368 0.846 0.805 0.892 0.299 0.758 5.200 24.600
KNN 0.629 0.339 0.476 0.263 0.837 0.781 0.901 0.206 0.738 4.200 25.600
MLP 0.621 0.432 0.444 0.421 0.812 0.805 0.820 0.245 0.718 7.200 22.600
XGBoost 0.610 0.415 0.386 0.447 0.778 0.800 0.757 0.195 0.678 8.800 21.000
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
NaiveBayes 96 15 24 14
LogisticRegression 86 25 14 24
SVM 84 27 14 24
DecisionTree 90 21 18 20
RandomForest 99 12 24 14
KNN 100 11 28 10
MLP 91 20 22 16
XGBoost 84 27 21 17
WCST_PCT_NONPERS_ERR_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'WCST_PCT_NONPERS_ERR_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
LogisticRegression 0.729 0.552 0.490 0.632 0.815 0.860 0.775 0.377 0.738 9.800 20.000
NaiveBayes 0.725 0.478 0.552 0.421 0.848 0.817 0.883 0.335 0.765 5.800 24.000
SVM 0.683 0.535 0.479 0.605 0.811 0.851 0.775 0.354 0.732 9.600 20.200
RandomForest 0.630 0.438 0.538 0.368 0.846 0.805 0.892 0.299 0.758 5.200 24.600
MLP 0.624 0.420 0.395 0.447 0.783 0.802 0.766 0.205 0.685 8.600 21.200
KNN 0.612 0.316 0.474 0.237 0.838 0.777 0.910 0.192 0.738 3.800 26.000
XGBoost 0.606 0.444 0.471 0.421 0.823 0.809 0.838 0.269 0.732 6.800 23.000
DecisionTree 0.567 0.351 0.361 0.342 0.786 0.779 0.793 0.137 0.678 7.200 22.600
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
LogisticRegression 86 25 14 24
NaiveBayes 98 13 22 16
SVM 86 25 15 23
RandomForest 99 12 24 14
MLP 85 26 21 17
KNN 101 10 29 9
XGBoost 93 18 22 16
DecisionTree 88 23 25 13
WCST_PCT_CONCEPTUAL_T
[資料] 來源=isi_raw_data 目標=psqi_lcga_group_2 列數=149 特徵數=3
[特徵] 使用欄位（前 15）：['BDI_T1', 'BAI_T1', 'WCST_PCT_CONCEPTUAL_T']
[CV] Stratified 5-fold, seed=42 | class_weight=balanced
[聚合] K-fold = weighted | LOSO = weighted
[Leakage check] Class balance
count percent%
psqi_lcga_group_2
0 111 74.5
1 38 25.5
[Leakage check] 未發現與目標 |r| ≥ 0.95 的欄位。
=== Basic ML Benchmark (Stratified 5-fold CV) ===
model AUC F1_pos(=1) Prec_pos Rec_pos F1_neg(=0) Prec_neg Rec_neg MCC Accuracy Pred1_mean Pred0_mean
NaiveBayes 0.734 0.441 0.500 0.395 0.835 0.807 0.865 0.282 0.745 6.000 23.800
LogisticRegression 0.727 0.545 0.480 0.632 0.810 0.859 0.766 0.367 0.732 10.000 19.800
RandomForest 0.707 0.373 0.524 0.289 0.845 0.789 0.910 0.250 0.752 4.200 25.600
KNN 0.686 0.300 0.409 0.237 0.824 0.772 0.883 0.147 0.718 4.400 25.400
SVM 0.683 0.562 0.466 0.711 0.792 0.879 0.721 0.386 0.718 11.600 18.200
XGBoost 0.654 0.410 0.400 0.421 0.791 0.798 0.784 0.201 0.691 8.000 21.800
DecisionTree 0.647 0.472 0.500 0.447 0.832 0.817 0.847 0.306 0.745 6.800 23.000
MLP 0.644 0.415 0.386 0.447 0.778 0.800 0.757 0.195 0.678 8.800 21.000
--- Aggregated Confusion Matrix Sums (across all folds' test parts) ---
model TN_sum FP_sum FN_sum TP_sum
NaiveBayes 96 15 23 15
LogisticRegression 85 26 14 24
RandomForest 101 10 27 11
KNN 98 13 29 9
SVM 80 31 11 27
XGBoost 87 24 22 16
DecisionTree 94 17 21 17
MLP 84 27 21 17