Model Details ============= H2OGradientBoostingEstimator : Gradient Boosting Machine Model Key: GBM_grid__1_AutoML_20200427_192307_model_1 Model Summary: number_of_trees number_of_internal_trees model_size_in_bytes min_depth max_depth mean_depth min_leaves max_leaves mean_leaves -- ----------------- -------------------------- --------------------- ----------- ----------- ------------ ------------ ------------ ------------- 111 111 351224 10 10 10 86 439 247.144 ModelMetricsBinomial: gbm ** Reported on train data. ** MSE: 0.0567336568895256 RMSE: 0.23818828033621972 LogLoss: 0.18829163422460377 Mean Per-Class Error: 0.09089018793422177 AUC: 0.9736733161182054 AUCPR: 0.9305432936941168 Gini: 0.9473466322364108 Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.41015156873835046: 0 1 Error Rate ----- ----- ---- ------- ---------------- 0 23395 1325 0.0536 (1325.0/24720.0) 1 1161 6680 0.1481 (1161.0/7841.0) Total 24556 8005 0.0763 (2486.0/32561.0) Maximum Metrics: Maximum metrics at their respective thresholds metric threshold value idx --------------------------- ----------- -------- ----- max f1 0.410152 0.843115 195 max f2 0.221135 0.883345 265 max f0point5 0.586657 0.873696 138 max accuracy 0.463723 0.924603 177 max precision 0.99812 1 0 max recall 0.00456462 1 391 max specificity 0.99812 1 0 max absolute_mcc 0.410152 0.792746 195 max min_per_class_accuracy 0.318646 0.906772 228 max mean_per_class_accuracy 0.287644 0.90911 240 max tns 0.99812 24720 0 max fns 0.99812 7659 0 max fps 0.000754229 24720 399 max tps 0.00456462 7841 391 max tnr 0.99812 1 0 max fnr 0.99812 0.976789 0 max fpr 0.000754229 1 399 max tpr 0.00456462 1 391 Gains/Lift Table: Avg response rate: 24,08 %, avg score: 24,10 % group cumulative_data_fraction lower_threshold lift cumulative_lift response_rate score cumulative_response_rate cumulative_score capture_rate cumulative_capture_rate gain cumulative_gain -- ------- -------------------------- ----------------- ---------- ----------------- --------------- ---------- -------------------------- ------------------ -------------- ------------------------- -------- ----------------- 1 0.010012 0.99697 4.15266 4.15266 1 0.997756 1 0.997756 0.0415763 0.0415763 315.266 315.266 2 0.020024 0.99529 4.15266 4.15266 1 0.996155 1 0.996955 0.0415763 0.0831527 315.266 315.266 3 0.0300052 0.992848 4.15266 4.15266 1 0.994155 1 0.996024 0.0414488 0.124601 315.266 315.266 4 0.0400172 0.988936 4.15266 4.15266 1 0.991113 1 0.994795 0.0415763 0.166178 315.266 315.266 5 0.0500292 0.981363 4.15266 4.15266 1 0.985503 1 0.992936 0.0415763 0.207754 315.266 315.266 6 0.100028 0.843866 4.1093 4.13098 0.989558 0.917575 0.99478 0.955267 0.205458 0.413213 310.93 313.098 7 0.150026 0.699372 3.72668 3.99624 0.89742 0.772147 0.962334 0.89424 0.186328 0.599541 272.668 299.624 8 0.200025 0.541961 3.00736 3.74906 0.724201 0.620315 0.90281 0.825769 0.150363 0.749904 200.736 274.906 9 0.300021 0.293837 1.70647 3.06827 0.410934 0.405798 0.738868 0.685793 0.170641 0.920546 70.6467 206.827 10 0.400018 0.131527 0.604533 2.45238 0.145577 0.204844 0.590557 0.565565 0.0604515 0.980997 -39.5467 145.238 11 0.500015 0.0531402 0.151771 1.99229 0.0365479 0.0870945 0.479762 0.469877 0.0151766 0.996174 -84.8229 99.2287 12 0.600012 0.0217344 0.0318847 1.66557 0.00767813 0.0349504 0.401085 0.397393 0.00318837 0.999362 -96.8115 66.557 13 0.700009 0.00954594 0.00510155 1.42837 0.0012285 0.0147878 0.343965 0.342737 0.000510139 0.999872 -99.4898 42.837 14 0.800006 0.00419856 0.00127539 1.24999 0.000307125 0.00655514 0.30101 0.300716 0.000127535 1 -99.8725 24.999 15 0.900003 0.00171594 0 1.11111 0 0.00276961 0.267565 0.267612 0 1 -100 11.1107 16 1 0.000162758 0 1 0 0.00112962 0.24081 0.240965 0 1 -100 0 ModelMetricsBinomial: gbm ** Reported on validation data. ** MSE: 0.09050010017572425 RMSE: 0.3008323456274678 LogLoss: 0.284810336314609 Mean Per-Class Error: 0.15920041626755532 AUC: 0.9240236541508303 AUCPR: 0.817208283334983 Gini: 0.8480473083016606 Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.37240703421420257: 0 1 Error Rate ----- ----- ---- ------- ---------------- 0 11142 1293 0.104 (1293.0/12435.0) 1 962 2884 0.2501 (962.0/3846.0) Total 12104 4177 0.1385 (2255.0/16281.0) Maximum Metrics: Maximum metrics at their respective thresholds metric threshold value idx --------------------------- ----------- -------- ----- max f1 0.372407 0.718933 207 max f2 0.12458 0.802609 305 max f0point5 0.637933 0.75522 120 max accuracy 0.523432 0.870155 155 max precision 0.998396 1 0 max recall 0.00107046 1 398 max specificity 0.998396 1 0 max absolute_mcc 0.372407 0.628187 207 max min_per_class_accuracy 0.243517 0.838118 256 max mean_per_class_accuracy 0.205008 0.8408 271 max tns 0.998396 12435 0 max fns 0.998396 3789 0 max fps 0.000724173 12435 399 max tps 0.00107046 3846 398 max tnr 0.998396 1 0 max fnr 0.998396 0.985179 0 max fpr 0.000724173 1 399 max tpr 0.00107046 1 398 Gains/Lift Table: Avg response rate: 23,62 %, avg score: 23,34 % group cumulative_data_fraction lower_threshold lift cumulative_lift response_rate score cumulative_response_rate cumulative_score capture_rate cumulative_capture_rate gain cumulative_gain -- ------- -------------------------- ----------------- ---------- ----------------- --------------- ---------- -------------------------- ------------------ -------------- ------------------------- -------- ----------------- 1 0.0100117 0.996763 4.23323 4.23323 1 0.997723 1 0.997723 0.0423817 0.0423817 323.323 323.323 2 0.0200233 0.994919 4.23323 4.23323 1 0.995925 1 0.996824 0.0423817 0.0847634 323.323 323.323 3 0.030035 0.991405 4.23323 4.23323 1 0.993272 1 0.99564 0.0423817 0.127145 323.323 323.323 4 0.0400467 0.986004 4.23323 4.23323 1 0.988901 1 0.993955 0.0423817 0.169527 323.323 323.323 5 0.0500584 0.973776 4.20726 4.22804 0.993865 0.980535 0.998773 0.991271 0.0421217 0.211648 320.726 322.804 6 0.100055 0.798633 3.57276 3.9006 0.84398 0.888854 0.921424 0.940094 0.178627 0.390276 257.276 290.06 7 0.150052 0.649405 2.88109 3.5609 0.68059 0.722431 0.841179 0.867569 0.144046 0.534321 188.109 256.09 8 0.200049 0.504334 2.23103 3.22854 0.527027 0.574928 0.762665 0.794432 0.111544 0.645866 123.103 222.854 9 0.300043 0.284855 1.67717 2.71152 0.396192 0.388324 0.640532 0.65909 0.167707 0.813573 67.717 171.152 10 0.400037 0.135753 1.0037 2.28463 0.237101 0.20382 0.53969 0.54529 0.100364 0.913937 0.370179 128.463 11 0.500031 0.0554428 0.499251 1.9276 0.117936 0.0899607 0.455349 0.454235 0.049922 0.963859 -50.0749 92.7599 12 0.600025 0.0230157 0.208021 1.64103 0.04914 0.0369824 0.387655 0.3847 0.0208008 0.984659 -79.1979 64.1032 13 0.700018 0.00971583 0.09881 1.42073 0.0233415 0.0154536 0.335615 0.331955 0.0098804 0.99454 -90.119 42.0734 14 0.800012 0.00427716 0.0416042 1.24836 0.00982801 0.00664879 0.294894 0.291295 0.00416017 0.9987 -95.8396 24.8356 15 0.900006 0.00170441 0.00780079 1.11053 0.00184275 0.00278255 0.262335 0.25924 0.000780031 0.99948 -99.2199 11.0526 16 1 0.000182443 0.00520053 1 0.0012285 0.00113032 0.236226 0.233431 0.000520021 1 -99.4799 0 ModelMetricsBinomial: gbm ** Reported on cross-validation data. ** MSE: 0.09186946368458541 RMSE: 0.3030997586349838 LogLoss: 0.28964576532382225 Mean Per-Class Error: 0.15938050870682652 AUC: 0.9235226940664146 AUCPR: 0.817936481363115 Gini: 0.8470453881328293 Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.30521033005951814: 0 1 Error Rate ----- ----- ---- ------- ---------------- 0 21435 3285 0.1329 (3285.0/24720.0) 1 1556 6285 0.1984 (1556.0/7841.0) Total 22991 9570 0.1487 (4841.0/32561.0) Maximum Metrics: Maximum metrics at their respective thresholds metric threshold value idx --------------------------- ----------- -------- ----- max f1 0.30521 0.721957 234 max f2 0.123447 0.804856 311 max f0point5 0.68143 0.751972 106 max accuracy 0.510241 0.868585 161 max precision 0.998228 1 0 max recall 0.00119949 1 398 max specificity 0.998228 1 0 max absolute_mcc 0.381258 0.630042 205 max min_per_class_accuracy 0.251431 0.839306 255 max mean_per_class_accuracy 0.21925 0.840619 267 max tns 0.998228 24720 0 max fns 0.998228 7683 0 max fps 0.000699384 24720 399 max tps 0.00119949 7841 398 max tnr 0.998228 1 0 max fnr 0.998228 0.97985 0 max fpr 0.000699384 1 399 max tpr 0.00119949 1 398 Gains/Lift Table: Avg response rate: 24,08 %, avg score: 23,80 % group cumulative_data_fraction lower_threshold lift cumulative_lift response_rate score cumulative_response_rate cumulative_score capture_rate cumulative_capture_rate gain cumulative_gain -- ------- -------------------------- ----------------- ---------- ----------------- --------------- ----------- -------------------------- ------------------ -------------- ------------------------- -------- ----------------- 1 0.010012 0.996911 4.15266 4.15266 1 0.99776 1 0.99776 0.0415763 0.0415763 315.266 315.266 2 0.020024 0.995047 4.15266 4.15266 1 0.996042 1 0.996901 0.0415763 0.0831527 315.266 315.266 3 0.0300052 0.992446 4.13988 4.14841 0.996923 0.993807 0.998976 0.995872 0.0413213 0.124474 313.988 314.841 4 0.0400172 0.987841 4.12718 4.1431 0.993865 0.990313 0.997698 0.994481 0.0413213 0.165795 312.718 314.31 5 0.0500292 0.979294 4.10171 4.13481 0.98773 0.984054 0.995703 0.992394 0.0410662 0.206861 310.171 313.481 6 0.100028 0.817225 3.52517 3.83009 0.848894 0.902561 0.922321 0.947491 0.176253 0.383114 252.517 283.009 7 0.150026 0.672626 2.82371 3.49469 0.679975 0.745737 0.841556 0.880254 0.141181 0.524295 182.371 249.469 8 0.200025 0.521466 2.34161 3.20647 0.563882 0.595302 0.772148 0.809027 0.117077 0.641372 134.161 220.647 9 0.300021 0.29381 1.68096 2.69802 0.404791 0.397854 0.649708 0.671983 0.168091 0.809463 68.096 169.802 10 0.400018 0.138834 1.00628 2.27512 0.242322 0.209859 0.547869 0.556461 0.100625 0.910088 0.628011 127.512 11 0.500015 0.0575005 0.524184 1.92495 0.126229 0.0928752 0.463546 0.46375 0.0524168 0.962505 -47.5816 92.495 12 0.600012 0.0223379 0.209163 1.639 0.0503686 0.0371877 0.394687 0.39266 0.0209157 0.98342 -79.0837 63.9001 13 0.700009 0.00920069 0.0918278 1.41799 0.022113 0.0146597 0.341464 0.338662 0.0091825 0.992603 -90.8172 41.7986 14 0.800006 0.00383832 0.0522909 1.24728 0.0125921 0.00618567 0.300357 0.297104 0.00522892 0.997832 -94.7709 24.728 15 0.900003 0.00147057 0.0204062 1.11097 0.004914 0.0024827 0.267531 0.264369 0.00204056 0.999872 -97.9594 11.0966 16 1 8.24963e-05 0.00127539 1 0.000307125 0.000936131 0.24081 0.238027 0.000127535 1 -99.8725 0 Cross-Validation Metrics Summary: mean sd cv_1_valid cv_2_valid cv_3_valid cv_4_valid cv_5_valid ----------- ---------- ------------ ------------ ------------ ------------ ------------ ------------ accuracy 0.85587025 0.0053450153 0.86427146 0.85626537 0.8548833 0.8495086 0.8544226 auc 0.9235996 0.0046755704 0.9299005 0.9252309 0.9170059 0.92233497 0.9235257 aucpr 0.8181381 0.01238913 0.838132 0.8171966 0.80506504 0.8116457 0.8186512 err 0.14412974 0.0053450153 0.13572854 0.14373465 0.1451167 0.1504914 0.1455774 err_count 938.6 34.753418 884.0 936.0 945.0 980.0 948.0 --- --- --- --- --- --- --- --- precision 0.6710671 0.020366764 0.70542634 0.6657366 0.66972476 0.6515151 0.6629328 r2 0.49720773 0.017367559 0.5235495 0.49698055 0.4779119 0.48614946 0.5014473 recall 0.7884873 0.022483995 0.78351784 0.77973855 0.7599219 0.79935896 0.81989926 rmse 0.30308446 0.0034180796 0.29875094 0.30069456 0.30682516 0.30595437 0.30319735 specificity 0.8771983 0.011480634 0.89113975 0.8797672 0.8842211 0.865307 0.8655565 See the whole table with table.as_data_frame() Scoring History: timestamp duration number_of_trees training_rmse training_logloss training_auc training_pr_auc training_lift training_classification_error validation_rmse validation_logloss validation_auc validation_pr_auc validation_lift validation_classification_error --- ------------------- ---------- ----------------- ------------------- ------------------- ------------------ ------------------- ----------------- ------------------------------- ------------------- -------------------- ------------------ ------------------- ----------------- --------------------------------- 2020-04-27 19:23:33 6.755 sec 0.0 0.42757492266114555 0.5520112931915784 0.5 0.24080955744602436 1.0 0.7591904425539756 0.4247875104215015 0.5467485574653612 0.5 0.23622627602727106 1.0 0.7637737239727289 2020-04-27 19:23:34 6.821 sec 5.0 0.3548464355040544 0.40802811178339243 0.929306395124953 0.8269551757582684 4.139920881507695 0.13092349743558246 0.35793665155901877 0.41354945042563185 0.9122548118651727 0.7869791731795631 4.20725859709235 0.1456298753147841 2020-04-27 19:23:34 6.919 sec 10.0 0.31965583911183837 0.34285832974679875 0.9361748483925462 0.8386461634332205 4.152659099604642 0.12975645711126807 0.32792013962040656 0.3560940244087162 0.9169066038877984 0.7956850293462011 4.233229329173167 0.14034764449358147 2020-04-27 19:23:34 7.017 sec 15.0 0.3027412183720379 0.30913213345221985 0.9391825042955273 0.8481939675268768 4.152659099604642 0.11980590276711403 0.3146718797186697 0.327880440405877 0.9188983860118377 0.8032382218279127 4.233229329173167 0.14348013021313188 2020-04-27 19:23:34 7.107 sec 20.0 0.29206161481747206 0.2861105683068164 0.9421428557425102 0.8559480654149333 4.152659099604642 0.12051226927919904 0.3077139078439141 0.3105545524754504 0.9204679309005895 0.8076226448995711 4.20725859709235 0.14280449603832687 --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- 2020-04-27 19:23:35 8.552 sec 95.0 0.24468972881019319 0.19737893127416908 0.9702739371175246 0.9221349400314159 4.152659099604642 0.08279843985135592 0.2998317724699488 0.28289578814234806 0.9247924360078545 0.8188466300364046 4.233229329173167 0.13543394140409065 2020-04-27 19:23:35 8.641 sec 100.0 0.24198969660397804 0.1936264068605919 0.9717085663731717 0.9257861479223982 4.152659099604642 0.08098645618992045 0.3002740594544993 0.28373176754035817 0.9245092369034529 0.8179720386714714 4.233229329173167 0.13948774645292059 2020-04-27 19:23:35 8.728 sec 105.0 0.24082370722622137 0.19183755331618813 0.972331853785739 0.9272565903634076 4.152659099604642 0.07883664506618347 0.3003947902329099 0.2839339594401493 0.9243949766032459 0.8179536223926899 4.233229329173167 0.13746084392850563 2020-04-27 19:23:36 8.851 sec 110.0 0.238765345337001 0.18902639044072667 0.973392388837366 0.9298727218206383 4.152659099604642 0.0767482571174104 0.3007632528442623 0.28469210473808676 0.9240760953317103 0.817337331760285 4.233229329173167 0.14556845402616547 2020-04-27 19:23:36 8.915 sec 111.0 0.23818828033621972 0.18829163422460377 0.9736733161182054 0.9305432936941168 4.152659099604642 0.07634900648014496 0.3008323456274678 0.284810336314609 0.9240236541508303 0.817208283334983 4.233229329173167 0.13850500583502243 See the whole table with table.as_data_frame() Variable Importances: variable relative_importance scaled_importance percentage ---------------------------- --------------------- ------------------- ------------ label_encoded_relationship 4219.6 1 0.264301 capital_gain 2447.56 0.580046 0.153307 label_encoded_occupation 1764.28 0.418115 0.110508 education_num 1358.69 0.321995 0.0851039 age 1252.9 0.296923 0.0784773 fnlwgt 1114.47 0.264118 0.0698066 hours_per_week 887.418 0.210309 0.0555849 label_encoded_marital_status 804.967 0.190769 0.0504204 capital_loss 615.867 0.145954 0.0385758 label_encoded_education 602.978 0.142899 0.0377685 label_encoded_workclass 426.43 0.101059 0.0267101 label_encoded_native_country 212.988 0.0504759 0.0133408 label_encoded_sex 152.951 0.0362478 0.00958034 label_encoded_race 104.005 0.0246481 0.00651453