more script consolidations

1 files changed, 262 insertions, 217 deletions

diff --git a/data.yaml b/data.yaml
index a7bbe6e..360899a 100644
--- a/data.yaml
+++ b/data.yaml
@@ -1,11 +1,11 @@
 ---
 :cv:
-  lazar-mp2d-all:
-    :tp: 3334
-    :fp: 830
-    :tn: 3054
-    :fn: 572
-    :n: 7790
+  mp2d_lazar_all:
+    :tp: 3327
+    :fp: 827
+    :tn: 3049
+    :fn: 574
+    :n: 7777
     :acc: 0.82
     :tpr: 0.85
     :fpr: 0.21
@@ -17,182 +17,182 @@
     :tnr_perc: 79
     :ppv_perc: 80
     :npv_perc: 84
-  lazar-mp2d-high-confidence:
-    :tp: 2807
-    :fp: 560
-    :tn: 2089
-    :fn: 352
-    :n: 5808
+  mp2d_lazar_high_confidence:
+    :tp: 2959
+    :fp: 639
+    :tn: 2302
+    :fn: 400
+    :n: 6300
     :acc: 0.84
-    :tpr: 0.89
-    :fpr: 0.21
-    :tnr: 0.79
-    :ppv: 0.83
-    :npv: 0.86
+    :tpr: 0.88
+    :fpr: 0.22
+    :tnr: 0.78
+    :ppv: 0.82
+    :npv: 0.85
     :acc_perc: 84
-    :tpr_perc: 89
-    :tnr_perc: 79
-    :ppv_perc: 83
-    :npv_perc: 86
-  lazar-cdk-all:
-    :tp: 700
-    :fp: 1583
-    :tn: 2410
-    :fn: 3390
-    :n: 8083
-    :acc: 0.38
-    :tpr: 0.17
-    :fpr: 0.4
-    :tnr: 0.6
-    :ppv: 0.31
-    :npv: 0.42
-    :acc_perc: 38
-    :tpr_perc: 17
-    :tnr_perc: 60
-    :ppv_perc: 31
-    :npv_perc: 42
-  lazar-cdk-high-confidence:
-    :tp: 700
-    :fp: 1583
-    :tn: 2410
-    :fn: 3390
-    :n: 8083
-    :acc: 0.38
-    :tpr: 0.17
-    :fpr: 0.4
-    :tnr: 0.6
-    :ppv: 0.31
-    :npv: 0.42
-    :acc_perc: 38
-    :tpr_perc: 17
-    :tnr_perc: 60
-    :ppv_perc: 31
-    :npv_perc: 42
-  tensorflow-rf-mp2d:
-    :tp: 3197
-    :fp: 893
-    :tn: 3458
-    :fn: 742
-    :n: 8290
+    :tpr_perc: 88
+    :tnr_perc: 78
+    :ppv_perc: 82
+    :npv_perc: 85
+  mp2d_rf:
+    :tp: 3196
+    :fp: 749
+    :tn: 3467
+    :fn: 891
+    :n: 8303
     :acc: 0.8
-    :tpr: 0.81
-    :fpr: 0.21
-    :tnr: 0.79
-    :ppv: 0.78
-    :npv: 0.82
+    :tpr: 0.78
+    :fpr: 0.18
+    :tnr: 0.82
+    :ppv: 0.81
+    :npv: 0.8
     :acc_perc: 80
-    :tpr_perc: 81
-    :tnr_perc: 79
-    :ppv_perc: 78
-    :npv_perc: 82
-  tensorflow-lr-mp2d:
-    :tp: 3414
-    :fp: 676
-    :tn: 3541
-    :fn: 659
-    :n: 8290
+    :tpr_perc: 78
+    :tnr_perc: 82
+    :ppv_perc: 81
+    :npv_perc: 80
+  mp2d_lr:
+    :tp: 3409
+    :fp: 666
+    :tn: 3550
+    :fn: 678
+    :n: 8303
     :acc: 0.84
-    :tpr: 0.84
+    :tpr: 0.83
     :fpr: 0.16
     :tnr: 0.84
-    :ppv: 0.83
+    :ppv: 0.84
     :npv: 0.84
     :acc_perc: 84
-    :tpr_perc: 84
+    :tpr_perc: 83
     :tnr_perc: 84
-    :ppv_perc: 83
+    :ppv_perc: 84
     :npv_perc: 84
-  tensorflow-lr2-mp2d:
-    :tp: 3398
-    :fp: 692
-    :tn: 3560
-    :fn: 640
-    :n: 8290
+  mp2d_lr2:
+    :tp: 3392
+    :fp: 647
+    :tn: 3569
+    :fn: 695
+    :n: 8303
     :acc: 0.84
-    :tpr: 0.84
-    :fpr: 0.16
-    :tnr: 0.84
-    :ppv: 0.83
-    :npv: 0.85
+    :tpr: 0.83
+    :fpr: 0.15
+    :tnr: 0.85
+    :ppv: 0.84
+    :npv: 0.84
     :acc_perc: 84
-    :tpr_perc: 84
-    :tnr_perc: 84
-    :ppv_perc: 83
-    :npv_perc: 85
-  tensorflow-nn-mp2d:
-    :tp: 3375
-    :fp: 715
-    :tn: 3585
-    :fn: 615
-    :n: 8290
+    :tpr_perc: 83
+    :tnr_perc: 85
+    :ppv_perc: 84
+    :npv_perc: 84
+  mp2d_nn:
+    :tp: 3370
+    :fp: 623
+    :tn: 3593
+    :fn: 717
+    :n: 8303
     :acc: 0.84
-    :tpr: 0.85
-    :fpr: 0.17
-    :tnr: 0.83
-    :ppv: 0.83
-    :npv: 0.85
+    :tpr: 0.82
+    :fpr: 0.15
+    :tnr: 0.85
+    :ppv: 0.84
+    :npv: 0.83
     :acc_perc: 84
-    :tpr_perc: 85
-    :tnr_perc: 83
-    :ppv_perc: 83
-    :npv_perc: 85
-  tensorflow-svm-mp2d:
-    :tp: 3390
-    :fp: 700
-    :tn: 3611
-    :fn: 589
-    :n: 8290
+    :tpr_perc: 82
+    :tnr_perc: 85
+    :ppv_perc: 84
+    :npv_perc: 83
+  mp2d_svm:
+    :tp: 3384
+    :fp: 599
+    :tn: 3617
+    :fn: 703
+    :n: 8303
     :acc: 0.84
-    :tpr: 0.85
-    :fpr: 0.16
-    :tnr: 0.84
-    :ppv: 0.83
-    :npv: 0.86
+    :tpr: 0.83
+    :fpr: 0.14
+    :tnr: 0.86
+    :ppv: 0.85
+    :npv: 0.84
     :acc_perc: 84
-    :tpr_perc: 85
-    :tnr_perc: 84
-    :ppv_perc: 83
-    :npv_perc: 86
-  tensorflow-rf-cdk:
-    :tp: 3273
-    :fp: 747
-    :tn: 3483
-    :fn: 562
-    :n: 8065
+    :tpr_perc: 83
+    :tnr_perc: 86
+    :ppv_perc: 85
+    :npv_perc: 84
+  cdk_lazar_all:
+    :tp: 370
+    :fp: 345
+    :tn: 54
+    :fn: 42
+    :n: 811
+    :acc: 0.52
+    :tpr: 0.9
+    :fpr: 0.86
+    :tnr: 0.14
+    :ppv: 0.52
+    :npv: 0.56
+    :acc_perc: 52
+    :tpr_perc: 90
+    :tnr_perc: 14
+    :ppv_perc: 52
+    :npv_perc: 56
+  cdk_lazar_high_confidence:
+    :tp: 370
+    :fp: 345
+    :tn: 54
+    :fn: 42
+    :n: 811
+    :acc: 0.52
+    :tpr: 0.9
+    :fpr: 0.86
+    :tnr: 0.14
+    :ppv: 0.52
+    :npv: 0.56
+    :acc_perc: 52
+    :tpr_perc: 90
+    :tnr_perc: 14
+    :ppv_perc: 52
+    :npv_perc: 56
+  cdk_rf:
+    :tp: 3267
+    :fp: 568
+    :tn: 3492
+    :fn: 750
+    :n: 8077
     :acc: 0.84
-    :tpr: 0.85
-    :fpr: 0.18
-    :tnr: 0.82
-    :ppv: 0.81
-    :npv: 0.86
+    :tpr: 0.81
+    :fpr: 0.14
+    :tnr: 0.86
+    :ppv: 0.85
+    :npv: 0.82
     :acc_perc: 84
-    :tpr_perc: 85
-    :tnr_perc: 82
-    :ppv_perc: 81
-    :npv_perc: 86
-  tensorflow-lr-cdk:
-    :tp: 3252
-    :fp: 768
-    :tn: 3165
-    :fn: 880
-    :n: 8065
-    :acc: 0.8
-    :tpr: 0.79
-    :fpr: 0.2
-    :tnr: 0.8
-    :ppv: 0.81
-    :npv: 0.78
-    :acc_perc: 80
-    :tpr_perc: 79
-    :tnr_perc: 80
-    :ppv_perc: 81
-    :npv_perc: 78
-  tensorflow-lr2-cdk:
-    :tp: 3215
-    :fp: 805
-    :tn: 3251
-    :fn: 794
-    :n: 8065
+    :tpr_perc: 81
+    :tnr_perc: 86
+    :ppv_perc: 85
+    :npv_perc: 82
+  cdk_lr:
+    :tp: 3247
+    :fp: 887
+    :tn: 3173
+    :fn: 770
+    :n: 8077
+    :acc: 0.79
+    :tpr: 0.81
+    :fpr: 0.22
+    :tnr: 0.78
+    :ppv: 0.79
+    :npv: 0.8
+    :acc_perc: 79
+    :tpr_perc: 81
+    :tnr_perc: 78
+    :ppv_perc: 79
+    :npv_perc: 80
+  cdk_lr2:
+    :tp: 3210
+    :fp: 803
+    :tn: 3257
+    :fn: 807
+    :n: 8077
     :acc: 0.8
     :tpr: 0.8
     :fpr: 0.2
@@ -204,12 +204,12 @@
     :tnr_perc: 80
     :ppv_perc: 80
     :npv_perc: 80
-  tensorflow-nn-cdk:
-    :tp: 3419
-    :fp: 601
-    :tn: 3435
-    :fn: 610
-    :n: 8065
+  cdk_nn:
+    :tp: 3414
+    :fp: 617
+    :tn: 3443
+    :fn: 603
+    :n: 8077
     :acc: 0.85
     :tpr: 0.85
     :fpr: 0.15
@@ -221,12 +221,12 @@
     :tnr_perc: 85
     :ppv_perc: 85
     :npv_perc: 85
-  tensorflow-svm-cdk:
-    :tp: 3292
-    :fp: 728
-    :tn: 3332
-    :fn: 713
-    :n: 8065
+  cdk_svm:
+    :tp: 3286
+    :fp: 720
+    :tn: 3340
+    :fn: 731
+    :n: 8077
     :acc: 0.82
     :tpr: 0.82
     :fpr: 0.18
@@ -245,120 +245,165 @@
     :n_compounds: 8083
 :pa:
   :n: 602
-  :lazar_all_mp2d:
+  :mp2d_lazar_all:
     :n: 560
     :mut: 111
     :non_mut: 449
     :n_perc: 93
     :mut_perc: 20
     :non_mut_perc: 80
-  :lazar_high_confidence_mp2d:
+  :mp2d_lazar_high_confidence:
     :n: 301
     :mut: 76
     :non_mut: 225
     :n_perc: 50
     :mut_perc: 25
     :non_mut_perc: 75
-  :lazar_all_cdk:
-    :n: 560
-    :mut: 111
-    :non_mut: 449
-    :n_perc: 93
-    :mut_perc: 20
-    :non_mut_perc: 80
-  :lazar_high_confidence_cdk:
-    :n: 301
-    :mut: 76
-    :non_mut: 225
-    :n_perc: 50
-    :mut_perc: 25
-    :non_mut_perc: 75
-  :lr_sgd_mp2d:
+  :mp2d_rf:
+    :n: 602
+    :mut: 28
+    :non_mut: 574
+    :n_perc: 100
+    :mut_perc: 5
+    :non_mut_perc: 95
+  :mp2d_lr:
     :n: 602
     :mut: 127
     :non_mut: 475
     :n_perc: 100
     :mut_perc: 21
     :non_mut_perc: 79
-  :lr_scikit_mp2d:
+  :mp2d_lr2:
     :n: 602
     :mut: 118
     :non_mut: 484
     :n_perc: 100
     :mut_perc: 20
     :non_mut_perc: 80
-  :nn_mp2d:
+  :mp2d_nn:
     :n: 602
     :mut: 124
     :non_mut: 478
     :n_perc: 100
     :mut_perc: 21
     :non_mut_perc: 79
-  :rf_mp2d:
+  :mp2d_svm:
     :n: 602
-    :mut: 28
-    :non_mut: 574
+    :mut: 82
+    :non_mut: 520
     :n_perc: 100
-    :mut_perc: 5
-    :non_mut_perc: 95
-  :svm_mp2d:
+    :mut_perc: 14
+    :non_mut_perc: 86
+  :cdk_lazar_all:
+    :n: 560
+    :mut: 111
+    :non_mut: 449
+    :n_perc: 93
+    :mut_perc: 20
+    :non_mut_perc: 80
+  :cdk_lazar_high_confidence:
+    :n: 301
+    :mut: 76
+    :non_mut: 225
+    :n_perc: 50
+    :mut_perc: 25
+    :non_mut_perc: 75
+  :cdk_rf:
     :n: 602
-    :mut: 78
-    :non_mut: 524
+    :mut: 10
+    :non_mut: 592
     :n_perc: 100
-    :mut_perc: 13
-    :non_mut_perc: 87
-  :lr_sgd_cdk:
+    :mut_perc: 2
+    :non_mut_perc: 98
+  :cdk_lr:
     :n: 602
     :mut: 97
     :non_mut: 505
     :n_perc: 100
     :mut_perc: 16
     :non_mut_perc: 84
-  :lr_scikit_cdk:
+  :cdk_lr2:
     :n: 602
     :mut: 88
     :non_mut: 514
     :n_perc: 100
     :mut_perc: 15
     :non_mut_perc: 85
-  :nn_cdk:
+  :cdk_nn:
     :n: 602
     :mut: 150
     :non_mut: 452
     :n_perc: 100
     :mut_perc: 25
     :non_mut_perc: 75
-  :rf_cdk:
-    :n: 602
-    :mut: 10
-    :non_mut: 592
-    :n_perc: 100
-    :mut_perc: 2
-    :non_mut_perc: 98
-  :svm_cdk:
+  :cdk_svm:
     :n: 602
-    :mut: 18
-    :non_mut: 584
+    :mut: 19
+    :non_mut: 583
     :n_perc: 100
     :mut_perc: 3
     :non_mut_perc: 97
   :groups:
     :Retronecine:
       :n: 392
+      :mut: 629
+      :non_mut: 4547
+      :n_pred: 5176
+      :mut_perc: 12
+      :non_mut_perc: 87
     :Otonecine:
       :n: 46
+      :mut: 458
+      :non_mut: 176
+      :n_pred: 634
+      :mut_perc: 72
+      :non_mut_perc: 27
     :Platynecine:
       :n: 140
+      :mut: 76
+      :non_mut: 1588
+      :n_pred: 1664
+      :mut_perc: 4
+      :non_mut_perc: 95
     :N_oxide:
       :n: 84
+      :mut: 27
+      :non_mut: 1017
+      :n_pred: 1044
+      :mut_perc: 2
+      :non_mut_perc: 97
     :Dehydropyrrolizidine:
       :n: 23
+      :mut: 54
+      :non_mut: 202
+      :n_pred: 256
+      :mut_perc: 21
+      :non_mut_perc: 78
     :Tertiary_PA:
       :n: 495
+      :mut: 1136
+      :non_mut: 5306
+      :n_pred: 6442
+      :mut_perc: 17
+      :non_mut_perc: 82
     :Macrocyclic_diester:
       :n: 255
+      :mut: 984
+      :non_mut: 2382
+      :n_pred: 3366
+      :mut_perc: 29
+      :non_mut_perc: 70
     :Monoester:
       :n: 154
+      :mut: 45
+      :non_mut: 1895
+      :n_pred: 1940
+      :mut_perc: 2
+      :non_mut_perc: 97
     :Diester:
       :n: 163
+      :mut: 155
+      :non_mut: 1927
+      :n_pred: 2082
+      :mut_perc: 7
+      :non_mut_perc: 92