diff --git a/CHANGELOG.md b/CHANGELOG.md
index cf626aa7..6f4949e8 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,17 @@ Unreleased
- Add `model_info` class to better capture model information.
- Test `/examples` Jupyter notebooks within normal test suite.
+v1.10.1 (2023-08-24)
+----------
+**Improvements**
+- Introduced ability to specify the target index of a binary model when creating score code.
+ - index can be specified in `pzmm.import_model.ImportModel.import_model()`
+ - Relevant examples updated to include target_index.
+
+**Bugfixes**
+- Reworked `write_score_code.py` to allow for proper execution of single line scoring.
+- Added template files for `assess_model_bias.py` to allow for proper execution
+
v1.10 (2023-08-31)
----------
**Improvements**
diff --git a/examples/pzmm_binary_classification_model_import.ipynb b/examples/pzmm_binary_classification_model_import.ipynb
index c68074b0..e540b703 100644
--- a/examples/pzmm_binary_classification_model_import.ipynb
+++ b/examples/pzmm_binary_classification_model_import.ipynb
@@ -814,7 +814,8 @@
" predict_method=[dtc.predict_proba, [int, int]], # What is the predict method and what does it return?\n",
" score_metrics=score_metrics, # What are the output variables?\n",
" overwrite_model=True, # Overwrite the model if it already exists?\n",
- " target_values=[\"1\", \"0\"], # What are the expected values of the target variable?\n",
+ " target_values=[\"0\", \"1\"], # What are the expected values of the target variable?\n",
+ " target_index=1, # What is the index of the target value in target_values?\n",
" model_file_name=prefix + \".pickle\", # How was the model file serialized?\n",
" missing_values=True # Does the data include missing values?\n",
" )\n",
diff --git a/examples/pzmm_h2o_model_import.ipynb b/examples/pzmm_h2o_model_import.ipynb
index 2609b51f..76982ef9 100644
--- a/examples/pzmm_h2o_model_import.ipynb
+++ b/examples/pzmm_h2o_model_import.ipynb
@@ -87,7 +87,9 @@
"outputs": [
{
"data": {
- "text/plain": "'3.38.0.4'"
+ "text/plain": [
+ "'3.40.0.3'"
+ ]
},
"execution_count": 3,
"metadata": {},
@@ -114,14 +116,116 @@
"name": "stdout",
"output_type": "stream",
"text": [
- "Checking whether there is an H2O instance running at http://localhost:54321 . connected.\n",
- "Warning: Your H2O cluster version is too old (5 months and 8 days)!Please download and install the latest version from http://h2o.ai/download/\n"
+ "Checking whether there is an H2O instance running at http://localhost:54321..... not found.\n",
+ "Attempting to start a local H2O server...\n",
+ " Java Version: openjdk version \"11.0.13\" 2021-10-19; OpenJDK Runtime Environment JBR-11.0.13.7-1751.21-jcef (build 11.0.13+7-b1751.21); OpenJDK 64-Bit Server VM JBR-11.0.13.7-1751.21-jcef (build 11.0.13+7-b1751.21, mixed mode)\n",
+ " Starting server from /Users/dalmoo/opt/anaconda3/envs/yeehaw/lib/python3.8/site-packages/h2o/backend/bin/h2o.jar\n",
+ " Ice root: /var/folders/vs/np2dp7cs1y7ggk5pl92q_rb40000gn/T/tmp0_92875_\n",
+ " JVM stdout: /var/folders/vs/np2dp7cs1y7ggk5pl92q_rb40000gn/T/tmp0_92875_/h2o_dalmoo_started_from_python.out\n",
+ " JVM stderr: /var/folders/vs/np2dp7cs1y7ggk5pl92q_rb40000gn/T/tmp0_92875_/h2o_dalmoo_started_from_python.err\n",
+ " Server is running at http://127.0.0.1:54321\n",
+ "Connecting to H2O server at http://127.0.0.1:54321 ... successful.\n",
+ "Warning: Your H2O cluster version is (6 months and 1 day) old. There may be a newer version available.\n",
+ "Please download and install the latest version from: https://h2o-release.s3.amazonaws.com/h2o/latest_stable.html\n"
]
},
{
"data": {
- "text/plain": "-------------------------- -----------------------------\nH2O_cluster_uptime: 1 day 4 hours 9 mins\nH2O_cluster_timezone: America/New_York\nH2O_data_parsing_timezone: UTC\nH2O_cluster_version: 3.38.0.4\nH2O_cluster_version_age: 5 months and 8 days !!!\nH2O_cluster_name: H2O_from_python_demo_uqabco\nH2O_cluster_total_nodes: 1\nH2O_cluster_free_memory: 15.93 Gb\nH2O_cluster_total_cores: 16\nH2O_cluster_allowed_cores: 16\nH2O_cluster_status: locked, healthy\nH2O_connection_url: http://localhost:54321\nH2O_connection_proxy: {\"http\": null, \"https\": null}\nH2O_internal_security: False\nPython_version: 3.8.16 final\n-------------------------- -----------------------------",
- "text/html": "\n \n\n
\n \n \n | H2O_cluster_uptime: | \n1 day 4 hours 9 mins |
\n| H2O_cluster_timezone: | \nAmerica/New_York |
\n| H2O_data_parsing_timezone: | \nUTC |
\n| H2O_cluster_version: | \n3.38.0.4 |
\n| H2O_cluster_version_age: | \n5 months and 8 days !!! |
\n| H2O_cluster_name: | \nH2O_from_python_demo_uqabco |
\n| H2O_cluster_total_nodes: | \n1 |
\n| H2O_cluster_free_memory: | \n15.93 Gb |
\n| H2O_cluster_total_cores: | \n16 |
\n| H2O_cluster_allowed_cores: | \n16 |
\n| H2O_cluster_status: | \nlocked, healthy |
\n| H2O_connection_url: | \nhttp://localhost:54321 |
\n| H2O_connection_proxy: | \n{\"http\": null, \"https\": null} |
\n| H2O_internal_security: | \nFalse |
\n| Python_version: | \n3.8.16 final |
\n
\n
\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | H2O_cluster_uptime: | \n",
+ "02 secs |
\n",
+ "| H2O_cluster_timezone: | \n",
+ "America/Chicago |
\n",
+ "| H2O_data_parsing_timezone: | \n",
+ "UTC |
\n",
+ "| H2O_cluster_version: | \n",
+ "3.40.0.3 |
\n",
+ "| H2O_cluster_version_age: | \n",
+ "6 months and 1 day |
\n",
+ "| H2O_cluster_name: | \n",
+ "H2O_from_python_dalmoo_6awy1u |
\n",
+ "| H2O_cluster_total_nodes: | \n",
+ "1 |
\n",
+ "| H2O_cluster_free_memory: | \n",
+ "4 Gb |
\n",
+ "| H2O_cluster_total_cores: | \n",
+ "10 |
\n",
+ "| H2O_cluster_allowed_cores: | \n",
+ "10 |
\n",
+ "| H2O_cluster_status: | \n",
+ "locked, healthy |
\n",
+ "| H2O_connection_url: | \n",
+ "http://127.0.0.1:54321 |
\n",
+ "| H2O_connection_proxy: | \n",
+ "{\"http\": null, \"https\": null} |
\n",
+ "| H2O_internal_security: | \n",
+ "False |
\n",
+ "| Python_version: | \n",
+ "3.8.16 final |
\n",
+ "
\n",
+ "
Model Details\n=============\nH2OGeneralizedLinearEstimator : Generalized Linear Modeling\nModel Key: glmfit\n
\n\n \n
\n
\n GLM Model: summary\n | \nfamily | \nlink | \nregularization | \nlambda_search | \nnumber_of_predictors_total | \nnumber_of_active_predictors | \nnumber_of_iterations | \ntraining_frame |
|---|
\n | \nbinomial | \nlogit | \nElastic Net (alpha = 0.5, lambda = 9.244E-4 ) | \nnlambda = 100, lambda.max = 0.2455, lambda.min = 9.244E-4, lambda.1se = -1.0 | \n18 | \n17 | \n93 | \npy_3_sid_80dd |
\n
\n
\n
ModelMetricsBinomialGLM: glm\n** Reported on train data. **\n\nMSE: 0.12410460050039399\nRMSE: 0.3522848286548741\nLogLoss: 0.40208193579455126\nAUC: 0.78494005994006\nAUCPR: 0.569122888911077\nGini: 0.56988011988012\nNull degrees of freedom: 3580\nResidual degrees of freedom: 3563\nNull deviance: 3597.150438148382\nResidual deviance: 2879.710824160576\nAIC: 2915.710824160576
\n
\n \n
\n
\n Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.2574175263537824\n | \n0 | \n1 | \nError | \nRate |
|---|
\n | 0 | \n2446.0 | \n414.0 | \n0.1448 | \n (414.0/2860.0) |
\n| 1 | \n314.0 | \n407.0 | \n0.4355 | \n (314.0/721.0) |
\n| Total | \n2760.0 | \n821.0 | \n0.2033 | \n (728.0/3581.0) |
\n
\n
\n
\n
\n \n
\n
\n Maximum Metrics: Maximum metrics at their respective thresholds\n | metric | \nthreshold | \nvalue | \nidx |
|---|
\n | max f1 | \n0.2574175 | \n0.5278859 | \n208.0 |
\n| max f2 | \n0.1522974 | \n0.6261121 | \n279.0 |
\n| max f0point5 | \n0.3886078 | \n0.5571256 | \n145.0 |
\n| max accuracy | \n0.5672639 | \n0.8352416 | \n91.0 |
\n| max precision | \n0.9988094 | \n1.0 | \n0.0 |
\n| max recall | \n0.0012376 | \n1.0 | \n399.0 |
\n| max specificity | \n0.9988094 | \n1.0 | \n0.0 |
\n| max absolute_mcc | \n0.2666152 | \n0.4020977 | \n204.0 |
\n| max min_per_class_accuracy | \n0.1782785 | \n0.7066434 | \n258.0 |
\n| max mean_per_class_accuracy | \n0.2143163 | \n0.7195026 | \n234.0 |
\n| max tns | \n0.9988094 | \n2860.0 | \n0.0 |
\n| max fns | \n0.9988094 | \n717.0 | \n0.0 |
\n| max fps | \n0.0062619 | \n2860.0 | \n398.0 |
\n| max tps | \n0.0012376 | \n721.0 | \n399.0 |
\n| max tnr | \n0.9988094 | \n1.0 | \n0.0 |
\n| max fnr | \n0.9988094 | \n0.9944521 | \n0.0 |
\n| max fpr | \n0.0062619 | \n1.0 | \n398.0 |
\n| max tpr | \n0.0012376 | \n1.0 | \n399.0 |
\n
\n
\n
\n
\n \n
\n
\n Gains/Lift Table: Avg response rate: 20.13 %, avg score: 20.13 %\n | group | \ncumulative_data_fraction | \nlower_threshold | \nlift | \ncumulative_lift | \nresponse_rate | \nscore | \ncumulative_response_rate | \ncumulative_score | \ncapture_rate | \ncumulative_capture_rate | \ngain | \ncumulative_gain | \nkolmogorov_smirnov |
|---|
\n | 1 | \n0.0100531 | \n0.9466854 | \n4.9667129 | \n4.9667129 | \n1.0 | \n0.9805079 | \n1.0 | \n0.9805079 | \n0.0499307 | \n0.0499307 | \n396.6712899 | \n396.6712899 | \n0.0499307 |
\n| 2 | \n0.0201061 | \n0.8523470 | \n4.8287487 | \n4.8977308 | \n0.9722222 | \n0.8987982 | \n0.9861111 | \n0.9396530 | \n0.0485437 | \n0.0984743 | \n382.8748652 | \n389.7730775 | \n0.0981247 |
\n| 3 | \n0.0301592 | \n0.7479588 | \n3.8629989 | \n4.5528202 | \n0.7777778 | \n0.7984125 | \n0.9166667 | \n0.8925729 | \n0.0388350 | \n0.1373093 | \n286.2998921 | \n355.2820157 | \n0.1341624 |
\n| 4 | \n0.0402122 | \n0.6912384 | \n3.1731777 | \n4.2079095 | \n0.6388889 | \n0.7152522 | \n0.8472222 | \n0.8482427 | \n0.0319001 | \n0.1692094 | \n217.3177685 | \n320.7909539 | \n0.1615171 |
\n| 5 | \n0.0502653 | \n0.6267417 | \n3.5870704 | \n4.0837417 | \n0.7222222 | \n0.6558977 | \n0.8222222 | \n0.8097737 | \n0.0360610 | \n0.2052705 | \n258.7070427 | \n308.3741717 | \n0.1940816 |
\n| 6 | \n0.1002513 | \n0.4302426 | \n2.4417360 | \n3.2650257 | \n0.4916201 | \n0.5199788 | \n0.6573816 | \n0.6652798 | \n0.1220527 | \n0.3273232 | \n144.1735950 | \n226.5025750 | \n0.2843162 |
\n| 7 | \n0.1502374 | \n0.3459143 | \n2.0532780 | \n2.8618606 | \n0.4134078 | \n0.3838592 | \n0.5762082 | \n0.5716473 | \n0.1026352 | \n0.4299584 | \n105.3277958 | \n186.1860592 | \n0.3502381 |
\n| 8 | \n0.2002234 | \n0.2844270 | \n1.7203140 | \n2.5768720 | \n0.3463687 | \n0.3119048 | \n0.5188285 | \n0.5068022 | \n0.0859917 | \n0.5159501 | \n72.0313965 | \n157.6871964 | \n0.3953207 |
\n| 9 | \n0.3001955 | \n0.2098447 | \n1.3318560 | \n2.1622527 | \n0.2681564 | \n0.2431157 | \n0.4353488 | \n0.4189885 | \n0.1331484 | \n0.6490985 | \n33.1855973 | \n116.2252685 | \n0.4368607 |
\n| 10 | \n0.4001676 | \n0.1695745 | \n0.8879040 | \n1.8438878 | \n0.1787709 | \n0.1882518 | \n0.3712491 | \n0.3613446 | \n0.0887656 | \n0.7378641 | \n-11.2096018 | \n84.3887831 | \n0.4228291 |
\n| 11 | \n0.5001396 | \n0.1380308 | \n0.6798015 | \n1.6112006 | \n0.1368715 | \n0.1525308 | \n0.3243998 | \n0.3196051 | \n0.0679612 | \n0.8058252 | \n-32.0198514 | \n61.1200555 | \n0.3827483 |
\n| 12 | \n0.6001117 | \n0.1146180 | \n0.5410665 | \n1.4329279 | \n0.1089385 | \n0.1258826 | \n0.2885063 | \n0.2873331 | \n0.0540915 | \n0.8599168 | \n-45.8933511 | \n43.2927872 | \n0.3253014 |
\n| 13 | \n0.7000838 | \n0.0945695 | \n0.5271930 | \n1.3035888 | \n0.1061453 | \n0.1047775 | \n0.2624651 | \n0.2612641 | \n0.0527046 | \n0.9126214 | \n-47.2807011 | \n30.3588786 | \n0.2661179 |
\n| 14 | \n0.8000559 | \n0.0736142 | \n0.3884580 | \n1.1892374 | \n0.0782123 | \n0.0841117 | \n0.2394415 | \n0.2391278 | \n0.0388350 | \n0.9514563 | \n-61.1542008 | \n18.9237364 | \n0.1895682 |
\n| 15 | \n0.9000279 | \n0.0521778 | \n0.2358495 | \n1.0833382 | \n0.0474860 | \n0.0627678 | \n0.2181198 | \n0.2195383 | \n0.0235784 | \n0.9750347 | \n-76.4150505 | \n8.3338246 | \n0.0939158 |
\n| 16 | \n1.0 | \n0.0011165 | \n0.2497230 | \n1.0 | \n0.0502793 | \n0.0375143 | \n0.2013404 | \n0.2013410 | \n0.0249653 | \n1.0 | \n-75.0277005 | \n0.0 | \n0.0 |
\n
\n
\n
ModelMetricsBinomialGLM: glm\n** Reported on validation data. **\n\nMSE: 0.11783296119533457\nRMSE: 0.3432680602609782\nLogLoss: 0.38278517425537495\nAUC: 0.8173716250592095\nAUCPR: 0.5982146215209964\nGini: 0.6347432501184189\nNull degrees of freedom: 1195\nResidual degrees of freedom: 1178\nNull deviance: 1193.6689319811862\nResidual deviance: 915.6221368188569\nAIC: 951.6221368188569
\n
\n \n
\n
\n Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.31271322943274704\n | \n0 | \n1 | \nError | \nRate |
|---|
\n | 0 | \n861.0 | \n97.0 | \n0.1013 | \n (97.0/958.0) |
\n| 1 | \n107.0 | \n131.0 | \n0.4496 | \n (107.0/238.0) |
\n| Total | \n968.0 | \n228.0 | \n0.1706 | \n (204.0/1196.0) |
\n
\n
\n
\n
\n \n
\n
\n Maximum Metrics: Maximum metrics at their respective thresholds\n | metric | \nthreshold | \nvalue | \nidx |
|---|
\n | max f1 | \n0.3127132 | \n0.5622318 | \n155.0 |
\n| max f2 | \n0.1877475 | \n0.6551476 | \n230.0 |
\n| max f0point5 | \n0.4296159 | \n0.6079404 | \n107.0 |
\n| max accuracy | \n0.4438566 | \n0.8461538 | \n105.0 |
\n| max precision | \n0.9939862 | \n1.0 | \n0.0 |
\n| max recall | \n0.0096573 | \n1.0 | \n398.0 |
\n| max specificity | \n0.9939862 | \n1.0 | \n0.0 |
\n| max absolute_mcc | \n0.3127132 | \n0.4565355 | \n155.0 |
\n| max min_per_class_accuracy | \n0.1954052 | \n0.7478992 | \n224.0 |
\n| max mean_per_class_accuracy | \n0.1914520 | \n0.7508026 | \n227.0 |
\n| max tns | \n0.9939862 | \n958.0 | \n0.0 |
\n| max fns | \n0.9939862 | \n237.0 | \n0.0 |
\n| max fps | \n0.0074970 | \n958.0 | \n399.0 |
\n| max tps | \n0.0096573 | \n238.0 | \n398.0 |
\n| max tnr | \n0.9939862 | \n1.0 | \n0.0 |
\n| max fnr | \n0.9939862 | \n0.9957983 | \n0.0 |
\n| max fpr | \n0.0074970 | \n1.0 | \n399.0 |
\n| max tpr | \n0.0096573 | \n1.0 | \n398.0 |
\n
\n
\n
\n
\n \n
\n
\n Gains/Lift Table: Avg response rate: 19.90 %, avg score: 21.55 %\n | group | \ncumulative_data_fraction | \nlower_threshold | \nlift | \ncumulative_lift | \nresponse_rate | \nscore | \ncumulative_response_rate | \ncumulative_score | \ncapture_rate | \ncumulative_capture_rate | \ngain | \ncumulative_gain | \nkolmogorov_smirnov |
|---|
\n | 1 | \n0.0100334 | \n0.9422870 | \n5.0252101 | \n5.0252101 | \n1.0 | \n0.9698172 | \n1.0 | \n0.9698172 | \n0.0504202 | \n0.0504202 | \n402.5210084 | \n402.5210084 | \n0.0504202 |
\n| 2 | \n0.0200669 | \n0.8859499 | \n4.6064426 | \n4.8158263 | \n0.9166667 | \n0.9133302 | \n0.9583333 | \n0.9415737 | \n0.0462185 | \n0.0966387 | \n360.6442577 | \n381.5826331 | \n0.0955948 |
\n| 3 | \n0.0301003 | \n0.8333951 | \n3.7689076 | \n4.4668534 | \n0.75 | \n0.8662565 | \n0.8888889 | \n0.9164679 | \n0.0378151 | \n0.1344538 | \n276.8907563 | \n346.6853408 | \n0.1302784 |
\n| 4 | \n0.0401338 | \n0.7848628 | \n3.7689076 | \n4.2923669 | \n0.75 | \n0.8154078 | \n0.8541667 | \n0.8912029 | \n0.0378151 | \n0.1722689 | \n276.8907563 | \n329.2366947 | \n0.1649620 |
\n| 5 | \n0.0501672 | \n0.7212019 | \n2.5126050 | \n3.9364146 | \n0.5 | \n0.7506003 | \n0.7833333 | \n0.8630824 | \n0.0252101 | \n0.1974790 | \n151.2605042 | \n293.6414566 | \n0.1839091 |
\n| 6 | \n0.1003344 | \n0.4947107 | \n3.2663866 | \n3.6014006 | \n0.65 | \n0.5984250 | \n0.7166667 | \n0.7307537 | \n0.1638655 | \n0.3613445 | \n226.6386555 | \n260.1400560 | \n0.3258539 |
\n| 7 | \n0.1505017 | \n0.3581892 | \n2.0100840 | \n3.0709617 | \n0.4 | \n0.4130834 | \n0.6111111 | \n0.6248636 | \n0.1008403 | \n0.4621849 | \n101.0084034 | \n207.0961718 | \n0.3891160 |
\n| 8 | \n0.2006689 | \n0.3002103 | \n2.0100840 | \n2.8057423 | \n0.4 | \n0.3274215 | \n0.5583333 | \n0.5505031 | \n0.1008403 | \n0.5630252 | \n101.0084034 | \n180.5742297 | \n0.4523780 |
\n| 9 | \n0.3001672 | \n0.2236898 | \n1.1824024 | \n2.2676435 | \n0.2352941 | \n0.2578008 | \n0.4512535 | \n0.4534792 | \n0.1176471 | \n0.6806723 | \n18.2402373 | \n126.7643548 | \n0.4750355 |
\n| 10 | \n0.4005017 | \n0.1761419 | \n1.0469188 | \n1.9618252 | \n0.2083333 | \n0.1963836 | \n0.3903967 | \n0.3890711 | \n0.1050420 | \n0.7857143 | \n4.6918768 | \n96.1825231 | \n0.4809126 |
\n| 11 | \n0.5 | \n0.1438627 | \n0.5489725 | \n1.6806723 | \n0.1092437 | \n0.1582716 | \n0.3344482 | \n0.3431428 | \n0.0546218 | \n0.8403361 | \n-45.1027470 | \n68.0672269 | \n0.4248873 |
\n| 12 | \n0.6003344 | \n0.1214392 | \n0.5025210 | \n1.4837668 | \n0.1 | \n0.1325910 | \n0.2952646 | \n0.3079531 | \n0.0504202 | \n0.8907563 | \n-49.7478992 | \n48.3766766 | \n0.3625726 |
\n| 13 | \n0.6998328 | \n0.0997811 | \n0.5067439 | \n1.3448591 | \n0.1008403 | \n0.1100543 | \n0.2676225 | \n0.2798169 | \n0.0504202 | \n0.9411765 | \n-49.3256126 | \n34.4859091 | \n0.3013017 |
\n| 14 | \n0.8001672 | \n0.0798711 | \n0.2512605 | \n1.2077307 | \n0.05 | \n0.0891164 | \n0.2403344 | \n0.2559046 | \n0.0252101 | \n0.9663866 | \n-74.8739496 | \n20.7730741 | \n0.2075139 |
\n| 15 | \n0.8996656 | \n0.0556177 | \n0.2111433 | \n1.0975134 | \n0.0420168 | \n0.0677358 | \n0.2184015 | \n0.2350941 | \n0.0210084 | \n0.9873950 | \n-78.8856719 | \n9.7513355 | \n0.1095244 |
\n| 16 | \n1.0 | \n0.0074970 | \n0.1256303 | \n1.0 | \n0.025 | \n0.0397561 | \n0.1989967 | \n0.2154950 | \n0.0126050 | \n1.0 | \n-87.4369748 | \n0.0 | \n0.0 |
\n
\n
\n
\n \n
\n
\n Scoring History: \n | \ntimestamp | \nduration | \niteration | \nlambda | \npredictors | \ndeviance_train | \ndeviance_test | \nalpha | \niterations | \ntraining_rmse | \ntraining_logloss | \ntraining_r2 | \ntraining_auc | \ntraining_pr_auc | \ntraining_lift | \ntraining_classification_error | \nvalidation_rmse | \nvalidation_logloss | \nvalidation_r2 | \nvalidation_auc | \nvalidation_pr_auc | \nvalidation_lift | \nvalidation_classification_error |
|---|
\n | \n2023-06-14 13:11:07 | \n 0.000 sec | \n1 | \n.25E0 | \n1 | \n1.0045100 | \n0.9980509 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.002 sec | \n3 | \n.22E0 | \n2 | \n0.9946355 | \n0.9864127 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.004 sec | \n5 | \n.2E0 | \n2 | \n0.9859035 | \n0.9760418 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.006 sec | \n7 | \n.19E0 | \n2 | \n0.9781697 | \n0.9667699 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.007 sec | \n9 | \n.17E0 | \n3 | \n0.9686784 | \n0.9544753 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.009 sec | \n11 | \n.15E0 | \n3 | \n0.9587676 | \n0.9411532 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.011 sec | \n13 | \n.14E0 | \n3 | \n0.9500953 | \n0.9293709 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.013 sec | \n15 | \n.13E0 | \n3 | \n0.9424603 | \n0.9188844 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.015 sec | \n17 | \n.12E0 | \n5 | \n0.9341431 | \n0.9080921 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.017 sec | \n19 | \n.11E0 | \n5 | \n0.9236449 | \n0.8955093 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n| --- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- | \n--- |
\n | \n2023-06-14 13:11:07 | \n 0.119 sec | \n89 | \n.13E-2 | \n18 | \n0.8045919 | \n0.7656444 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.120 sec | \n90 | \n.12E-2 | \n18 | \n0.8044580 | \n0.7656095 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.121 sec | \n91 | \n.11E-2 | \n18 | \n0.8043435 | \n0.7655877 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.122 sec | \n92 | \n.1E-2 | \n18 | \n0.8042461 | \n0.7655757 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.123 sec | \n93 | \n.92E-3 | \n18 | \n0.8041639 | \n0.7655703 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.125 sec | \n94 | \n.84E-3 | \n18 | \n0.8040939 | \n0.7655716 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.126 sec | \n95 | \n.77E-3 | \n18 | \n0.8040347 | \n0.7655772 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.127 sec | \n96 | \n.7E-3 | \n18 | \n0.8039850 | \n0.7655871 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.128 sec | \n97 | \n.64E-3 | \n18 | \n0.8039427 | \n0.7655987 | \n0.5 | \nNone | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n | \n |
\n | \n2023-06-14 13:11:07 | \n 0.130 sec | \n98 | \n.58E-3 | \n18 | \n0.8039071 | \n0.7656132 | \n0.5 | \n98 | \n0.3522848 | \n0.4020819 | \n0.2282170 | \n0.7849401 | \n0.5691229 | \n4.9667129 | \n0.2032952 | \n0.3432681 | \n0.3827852 | \n0.2607579 | \n0.8173716 | \n0.5982146 | \n5.0252101 | \n0.1705686 |
\n
\n
\n
[66 rows x 24 columns]
\n \n
\n
\n Variable Importances: \n | variable | \nrelative_importance | \nscaled_importance | \npercentage |
|---|
\n | DELINQ | \n0.7970149 | \n1.0 | \n0.1473006 |
\n| JOB.Sales | \n0.7478347 | \n0.9382946 | \n0.1382114 |
\n| JOB.Office | \n0.5109240 | \n0.6410470 | \n0.0944266 |
\n| JOB.Self | \n0.4571400 | \n0.5735652 | \n0.0844865 |
\n| CLAGE | \n0.4484982 | \n0.5627225 | \n0.0828894 |
\n| DEBTINC | \n0.4467601 | \n0.5605417 | \n0.0825681 |
\n| DEROG | \n0.4263238 | \n0.5349007 | \n0.0787912 |
\n| NINQ | \n0.3037353 | \n0.3810912 | \n0.0561350 |
\n| VALUE | \n0.2412667 | \n0.3027129 | \n0.0445898 |
\n| MORTDUE | \n0.2384536 | \n0.2991833 | \n0.0440699 |
\n| LOAN | \n0.2180902 | \n0.2736338 | \n0.0403064 |
\n| REASON.HomeImp | \n0.1862974 | \n0.2337439 | \n0.0344306 |
\n| CLNO | \n0.1485267 | \n0.1863538 | \n0.0274500 |
\n| REASON.DebtCon | \n0.1308125 | \n0.1641280 | \n0.0241762 |
\n| YOJ | \n0.0438346 | \n0.0549984 | \n0.0081013 |
\n| JOB.Other | \n0.0353206 | \n0.0443161 | \n0.0065278 |
\n| JOB.Mgr | \n0.0299719 | \n0.0376052 | \n0.0055393 |
\n| JOB.ProfExe | \n0.0 | \n0.0 | \n0.0 |
\n
\n
\n
\n\n[tips]\nUse `model.explain()` to inspect the model.\n--\nUse `h2o.display.toggle_user_tips()` to switch on/off this section.
"
+ "text/html": [
+ "Model Details\n",
+ "=============\n",
+ "H2OGeneralizedLinearEstimator : Generalized Linear Modeling\n",
+ "Model Key: glmfit\n",
+ "
\n",
+ "\n",
+ " \n",
+ "
\n",
+ "
\n",
+ " GLM Model: summary\n",
+ " | \n",
+ "family | \n",
+ "link | \n",
+ "regularization | \n",
+ "lambda_search | \n",
+ "number_of_predictors_total | \n",
+ "number_of_active_predictors | \n",
+ "number_of_iterations | \n",
+ "training_frame |
|---|
\n",
+ " | \n",
+ "binomial | \n",
+ "logit | \n",
+ "Elastic Net (alpha = 0.5, lambda = 9.244E-4 ) | \n",
+ "nlambda = 100, lambda.max = 0.2455, lambda.min = 9.244E-4, lambda.1se = -1.0 | \n",
+ "18 | \n",
+ "17 | \n",
+ "93 | \n",
+ "py_3_sid_a269 |
\n",
+ "
\n",
+ "
\n",
+ "
ModelMetricsBinomialGLM: glm\n",
+ "** Reported on train data. **\n",
+ "\n",
+ "MSE: 0.12410463474559494\n",
+ "RMSE: 0.3522848772592927\n",
+ "LogLoss: 0.40208197287665776\n",
+ "AUC: 0.7849415147958837\n",
+ "AUCPR: 0.5691248967099318\n",
+ "Gini: 0.5698830295917674\n",
+ "Null degrees of freedom: 3580\n",
+ "Residual degrees of freedom: 3563\n",
+ "Null deviance: 3597.150438148379\n",
+ "Residual deviance: 2879.711089742623\n",
+ "AIC: 2915.711089742623
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ " Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.2574171649920362\n",
+ " | \n",
+ "0 | \n",
+ "1 | \n",
+ "Error | \n",
+ "Rate |
|---|
\n",
+ " | 0 | \n",
+ "2446.0 | \n",
+ "414.0 | \n",
+ "0.1448 | \n",
+ " (414.0/2860.0) |
\n",
+ "| 1 | \n",
+ "314.0 | \n",
+ "407.0 | \n",
+ "0.4355 | \n",
+ " (314.0/721.0) |
\n",
+ "| Total | \n",
+ "2760.0 | \n",
+ "821.0 | \n",
+ "0.2033 | \n",
+ " (728.0/3581.0) |
\n",
+ "
\n",
+ "
\n",
+ "
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ " Maximum Metrics: Maximum metrics at their respective thresholds\n",
+ " | metric | \n",
+ "threshold | \n",
+ "value | \n",
+ "idx |
|---|
\n",
+ " | max f1 | \n",
+ "0.2574172 | \n",
+ "0.5278859 | \n",
+ "207.0 |
\n",
+ "| max f2 | \n",
+ "0.1522967 | \n",
+ "0.6261121 | \n",
+ "279.0 |
\n",
+ "| max f0point5 | \n",
+ "0.3886058 | \n",
+ "0.5571256 | \n",
+ "144.0 |
\n",
+ "| max accuracy | \n",
+ "0.5672592 | \n",
+ "0.8352416 | \n",
+ "90.0 |
\n",
+ "| max precision | \n",
+ "0.9988092 | \n",
+ "1.0 | \n",
+ "0.0 |
\n",
+ "| max recall | \n",
+ "0.0012377 | \n",
+ "1.0 | \n",
+ "399.0 |
\n",
+ "| max specificity | \n",
+ "0.9988092 | \n",
+ "1.0 | \n",
+ "0.0 |
\n",
+ "| max absolute_mcc | \n",
+ "0.2661852 | \n",
+ "0.4020977 | \n",
+ "203.0 |
\n",
+ "| max min_per_class_accuracy | \n",
+ "0.1782796 | \n",
+ "0.7066434 | \n",
+ "258.0 |
\n",
+ "| max mean_per_class_accuracy | \n",
+ "0.2143151 | \n",
+ "0.7195026 | \n",
+ "234.0 |
\n",
+ "| max tns | \n",
+ "0.9988092 | \n",
+ "2860.0 | \n",
+ "0.0 |
\n",
+ "| max fns | \n",
+ "0.9988092 | \n",
+ "717.0 | \n",
+ "0.0 |
\n",
+ "| max fps | \n",
+ "0.0062627 | \n",
+ "2860.0 | \n",
+ "398.0 |
\n",
+ "| max tps | \n",
+ "0.0012377 | \n",
+ "721.0 | \n",
+ "399.0 |
\n",
+ "| max tnr | \n",
+ "0.9988092 | \n",
+ "1.0 | \n",
+ "0.0 |
\n",
+ "| max fnr | \n",
+ "0.9988092 | \n",
+ "0.9944521 | \n",
+ "0.0 |
\n",
+ "| max fpr | \n",
+ "0.0062627 | \n",
+ "1.0 | \n",
+ "398.0 |
\n",
+ "| max tpr | \n",
+ "0.0012377 | \n",
+ "1.0 | \n",
+ "399.0 |
\n",
+ "
\n",
+ "
\n",
+ "
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ " Gains/Lift Table: Avg response rate: 20.13 %, avg score: 20.13 %\n",
+ " | group | \n",
+ "cumulative_data_fraction | \n",
+ "lower_threshold | \n",
+ "lift | \n",
+ "cumulative_lift | \n",
+ "response_rate | \n",
+ "score | \n",
+ "cumulative_response_rate | \n",
+ "cumulative_score | \n",
+ "capture_rate | \n",
+ "cumulative_capture_rate | \n",
+ "gain | \n",
+ "cumulative_gain | \n",
+ "kolmogorov_smirnov |
|---|
\n",
+ " | 1 | \n",
+ "0.0100531 | \n",
+ "0.9466824 | \n",
+ "4.9667129 | \n",
+ "4.9667129 | \n",
+ "1.0 | \n",
+ "0.9805069 | \n",
+ "1.0 | \n",
+ "0.9805069 | \n",
+ "0.0499307 | \n",
+ "0.0499307 | \n",
+ "396.6712899 | \n",
+ "396.6712899 | \n",
+ "0.0499307 |
\n",
+ "| 2 | \n",
+ "0.0201061 | \n",
+ "0.8523402 | \n",
+ "4.8287487 | \n",
+ "4.8977308 | \n",
+ "0.9722222 | \n",
+ "0.8987941 | \n",
+ "0.9861111 | \n",
+ "0.9396505 | \n",
+ "0.0485437 | \n",
+ "0.0984743 | \n",
+ "382.8748652 | \n",
+ "389.7730775 | \n",
+ "0.0981247 |
\n",
+ "| 3 | \n",
+ "0.0301592 | \n",
+ "0.7479558 | \n",
+ "3.8629989 | \n",
+ "4.5528202 | \n",
+ "0.7777778 | \n",
+ "0.7984072 | \n",
+ "0.9166667 | \n",
+ "0.8925694 | \n",
+ "0.0388350 | \n",
+ "0.1373093 | \n",
+ "286.2998921 | \n",
+ "355.2820157 | \n",
+ "0.1341624 |
\n",
+ "| 4 | \n",
+ "0.0402122 | \n",
+ "0.6912340 | \n",
+ "3.1731777 | \n",
+ "4.2079095 | \n",
+ "0.6388889 | \n",
+ "0.7152449 | \n",
+ "0.8472222 | \n",
+ "0.8482383 | \n",
+ "0.0319001 | \n",
+ "0.1692094 | \n",
+ "217.3177685 | \n",
+ "320.7909539 | \n",
+ "0.1615171 |
\n",
+ "| 5 | \n",
+ "0.0502653 | \n",
+ "0.6267302 | \n",
+ "3.5870704 | \n",
+ "4.0837417 | \n",
+ "0.7222222 | \n",
+ "0.6558936 | \n",
+ "0.8222222 | \n",
+ "0.8097693 | \n",
+ "0.0360610 | \n",
+ "0.2052705 | \n",
+ "258.7070427 | \n",
+ "308.3741717 | \n",
+ "0.1940816 |
\n",
+ "| 6 | \n",
+ "0.1002513 | \n",
+ "0.4302463 | \n",
+ "2.4417360 | \n",
+ "3.2650257 | \n",
+ "0.4916201 | \n",
+ "0.5199749 | \n",
+ "0.6573816 | \n",
+ "0.6652757 | \n",
+ "0.1220527 | \n",
+ "0.3273232 | \n",
+ "144.1735950 | \n",
+ "226.5025750 | \n",
+ "0.2843162 |
\n",
+ "| 7 | \n",
+ "0.1502374 | \n",
+ "0.3459178 | \n",
+ "2.0532780 | \n",
+ "2.8618606 | \n",
+ "0.4134078 | \n",
+ "0.3838564 | \n",
+ "0.5762082 | \n",
+ "0.5716436 | \n",
+ "0.1026352 | \n",
+ "0.4299584 | \n",
+ "105.3277958 | \n",
+ "186.1860592 | \n",
+ "0.3502381 |
\n",
+ "| 8 | \n",
+ "0.2002234 | \n",
+ "0.2844205 | \n",
+ "1.7203140 | \n",
+ "2.5768720 | \n",
+ "0.3463687 | \n",
+ "0.3119030 | \n",
+ "0.5188285 | \n",
+ "0.5067990 | \n",
+ "0.0859917 | \n",
+ "0.5159501 | \n",
+ "72.0313965 | \n",
+ "157.6871964 | \n",
+ "0.3953207 |
\n",
+ "| 9 | \n",
+ "0.3001955 | \n",
+ "0.2098441 | \n",
+ "1.3318560 | \n",
+ "2.1622527 | \n",
+ "0.2681564 | \n",
+ "0.2431147 | \n",
+ "0.4353488 | \n",
+ "0.4189860 | \n",
+ "0.1331484 | \n",
+ "0.6490985 | \n",
+ "33.1855973 | \n",
+ "116.2252685 | \n",
+ "0.4368607 |
\n",
+ "| 10 | \n",
+ "0.4001676 | \n",
+ "0.1695731 | \n",
+ "0.8879040 | \n",
+ "1.8438878 | \n",
+ "0.1787709 | \n",
+ "0.1882520 | \n",
+ "0.3712491 | \n",
+ "0.3613428 | \n",
+ "0.0887656 | \n",
+ "0.7378641 | \n",
+ "-11.2096018 | \n",
+ "84.3887831 | \n",
+ "0.4228291 |
\n",
+ "| 11 | \n",
+ "0.5001396 | \n",
+ "0.1380318 | \n",
+ "0.6798015 | \n",
+ "1.6112006 | \n",
+ "0.1368715 | \n",
+ "0.1525314 | \n",
+ "0.3243998 | \n",
+ "0.3196038 | \n",
+ "0.0679612 | \n",
+ "0.8058252 | \n",
+ "-32.0198514 | \n",
+ "61.1200555 | \n",
+ "0.3827483 |
\n",
+ "| 12 | \n",
+ "0.6001117 | \n",
+ "0.1146235 | \n",
+ "0.5410665 | \n",
+ "1.4329279 | \n",
+ "0.1089385 | \n",
+ "0.1258838 | \n",
+ "0.2885063 | \n",
+ "0.2873322 | \n",
+ "0.0540915 | \n",
+ "0.8599168 | \n",
+ "-45.8933511 | \n",
+ "43.2927872 | \n",
+ "0.3253014 |
\n",
+ "| 13 | \n",
+ "0.7000838 | \n",
+ "0.0945753 | \n",
+ "0.5271930 | \n",
+ "1.3035888 | \n",
+ "0.1061453 | \n",
+ "0.1047786 | \n",
+ "0.2624651 | \n",
+ "0.2612635 | \n",
+ "0.0527046 | \n",
+ "0.9126214 | \n",
+ "-47.2807011 | \n",
+ "30.3588786 | \n",
+ "0.2661179 |
\n",
+ "| 14 | \n",
+ "0.8000559 | \n",
+ "0.0736148 | \n",
+ "0.3884580 | \n",
+ "1.1892374 | \n",
+ "0.0782123 | \n",
+ "0.0841132 | \n",
+ "0.2394415 | \n",
+ "0.2391274 | \n",
+ "0.0388350 | \n",
+ "0.9514563 | \n",
+ "-61.1542008 | \n",
+ "18.9237364 | \n",
+ "0.1895682 |
\n",
+ "| 15 | \n",
+ "0.9000279 | \n",
+ "0.0521793 | \n",
+ "0.2358495 | \n",
+ "1.0833382 | \n",
+ "0.0474860 | \n",
+ "0.0627694 | \n",
+ "0.2181198 | \n",
+ "0.2195382 | \n",
+ "0.0235784 | \n",
+ "0.9750347 | \n",
+ "-76.4150505 | \n",
+ "8.3338246 | \n",
+ "0.0939158 |
\n",
+ "| 16 | \n",
+ "1.0 | \n",
+ "0.0011167 | \n",
+ "0.2497230 | \n",
+ "1.0 | \n",
+ "0.0502793 | \n",
+ "0.0375156 | \n",
+ "0.2013404 | \n",
+ "0.2013410 | \n",
+ "0.0249653 | \n",
+ "1.0 | \n",
+ "-75.0277005 | \n",
+ "0.0 | \n",
+ "0.0 |
\n",
+ "
\n",
+ "
\n",
+ "
ModelMetricsBinomialGLM: glm\n",
+ "** Reported on validation data. **\n",
+ "\n",
+ "MSE: 0.1178330502946334\n",
+ "RMSE: 0.3432681900418875\n",
+ "LogLoss: 0.3827854417227382\n",
+ "AUC: 0.8173891686110769\n",
+ "AUCPR: 0.5982392824442148\n",
+ "Gini: 0.6347783372221538\n",
+ "Null degrees of freedom: 1195\n",
+ "Residual degrees of freedom: 1178\n",
+ "Null deviance: 1193.6689319811862\n",
+ "Residual deviance: 915.6227766007898\n",
+ "AIC: 951.6227766007898
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ " Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.3129932533175332\n",
+ " | \n",
+ "0 | \n",
+ "1 | \n",
+ "Error | \n",
+ "Rate |
|---|
\n",
+ " | 0 | \n",
+ "863.0 | \n",
+ "95.0 | \n",
+ "0.0992 | \n",
+ " (95.0/958.0) |
\n",
+ "| 1 | \n",
+ "107.0 | \n",
+ "131.0 | \n",
+ "0.4496 | \n",
+ " (107.0/238.0) |
\n",
+ "| Total | \n",
+ "970.0 | \n",
+ "226.0 | \n",
+ "0.1689 | \n",
+ " (202.0/1196.0) |
\n",
+ "
\n",
+ "
\n",
+ "
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ " Maximum Metrics: Maximum metrics at their respective thresholds\n",
+ " | metric | \n",
+ "threshold | \n",
+ "value | \n",
+ "idx |
|---|
\n",
+ " | max f1 | \n",
+ "0.3129933 | \n",
+ "0.5646552 | \n",
+ "155.0 |
\n",
+ "| max f2 | \n",
+ "0.1877463 | \n",
+ "0.6551476 | \n",
+ "231.0 |
\n",
+ "| max f0point5 | \n",
+ "0.4296125 | \n",
+ "0.6079404 | \n",
+ "107.0 |
\n",
+ "| max accuracy | \n",
+ "0.4438485 | \n",
+ "0.8461538 | \n",
+ "105.0 |
\n",
+ "| max precision | \n",
+ "0.9939857 | \n",
+ "1.0 | \n",
+ "0.0 |
\n",
+ "| max recall | \n",
+ "0.0096581 | \n",
+ "1.0 | \n",
+ "398.0 |
\n",
+ "| max specificity | \n",
+ "0.9939857 | \n",
+ "1.0 | \n",
+ "0.0 |
\n",
+ "| max absolute_mcc | \n",
+ "0.3129933 | \n",
+ "0.4602072 | \n",
+ "155.0 |
\n",
+ "| max min_per_class_accuracy | \n",
+ "0.1954052 | \n",
+ "0.7478992 | \n",
+ "225.0 |
\n",
+ "| max mean_per_class_accuracy | \n",
+ "0.1914540 | \n",
+ "0.7508026 | \n",
+ "228.0 |
\n",
+ "| max tns | \n",
+ "0.9939857 | \n",
+ "958.0 | \n",
+ "0.0 |
\n",
+ "| max fns | \n",
+ "0.9939857 | \n",
+ "237.0 | \n",
+ "0.0 |
\n",
+ "| max fps | \n",
+ "0.0074980 | \n",
+ "958.0 | \n",
+ "399.0 |
\n",
+ "| max tps | \n",
+ "0.0096581 | \n",
+ "238.0 | \n",
+ "398.0 |
\n",
+ "| max tnr | \n",
+ "0.9939857 | \n",
+ "1.0 | \n",
+ "0.0 |
\n",
+ "| max fnr | \n",
+ "0.9939857 | \n",
+ "0.9957983 | \n",
+ "0.0 |
\n",
+ "| max fpr | \n",
+ "0.0074980 | \n",
+ "1.0 | \n",
+ "399.0 |
\n",
+ "| max tpr | \n",
+ "0.0096581 | \n",
+ "1.0 | \n",
+ "398.0 |
\n",
+ "
\n",
+ "
\n",
+ "
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ " Gains/Lift Table: Avg response rate: 19.90 %, avg score: 21.55 %\n",
+ " | group | \n",
+ "cumulative_data_fraction | \n",
+ "lower_threshold | \n",
+ "lift | \n",
+ "cumulative_lift | \n",
+ "response_rate | \n",
+ "score | \n",
+ "cumulative_response_rate | \n",
+ "cumulative_score | \n",
+ "capture_rate | \n",
+ "cumulative_capture_rate | \n",
+ "gain | \n",
+ "cumulative_gain | \n",
+ "kolmogorov_smirnov |
|---|
\n",
+ " | 1 | \n",
+ "0.0100334 | \n",
+ "0.9422827 | \n",
+ "5.0252101 | \n",
+ "5.0252101 | \n",
+ "1.0 | \n",
+ "0.9698159 | \n",
+ "1.0 | \n",
+ "0.9698159 | \n",
+ "0.0504202 | \n",
+ "0.0504202 | \n",
+ "402.5210084 | \n",
+ "402.5210084 | \n",
+ "0.0504202 |
\n",
+ "| 2 | \n",
+ "0.0200669 | \n",
+ "0.8859422 | \n",
+ "4.6064426 | \n",
+ "4.8158263 | \n",
+ "0.9166667 | \n",
+ "0.9133265 | \n",
+ "0.9583333 | \n",
+ "0.9415712 | \n",
+ "0.0462185 | \n",
+ "0.0966387 | \n",
+ "360.6442577 | \n",
+ "381.5826331 | \n",
+ "0.0955948 |
\n",
+ "| 3 | \n",
+ "0.0301003 | \n",
+ "0.8333880 | \n",
+ "3.7689076 | \n",
+ "4.4668534 | \n",
+ "0.75 | \n",
+ "0.8662506 | \n",
+ "0.8888889 | \n",
+ "0.9164643 | \n",
+ "0.0378151 | \n",
+ "0.1344538 | \n",
+ "276.8907563 | \n",
+ "346.6853408 | \n",
+ "0.1302784 |
\n",
+ "| 4 | \n",
+ "0.0401338 | \n",
+ "0.7848582 | \n",
+ "3.7689076 | \n",
+ "4.2923669 | \n",
+ "0.75 | \n",
+ "0.8154020 | \n",
+ "0.8541667 | \n",
+ "0.8911987 | \n",
+ "0.0378151 | \n",
+ "0.1722689 | \n",
+ "276.8907563 | \n",
+ "329.2366947 | \n",
+ "0.1649620 |
\n",
+ "| 5 | \n",
+ "0.0501672 | \n",
+ "0.7211959 | \n",
+ "2.5126050 | \n",
+ "3.9364146 | \n",
+ "0.5 | \n",
+ "0.7505935 | \n",
+ "0.7833333 | \n",
+ "0.8630777 | \n",
+ "0.0252101 | \n",
+ "0.1974790 | \n",
+ "151.2605042 | \n",
+ "293.6414566 | \n",
+ "0.1839091 |
\n",
+ "| 6 | \n",
+ "0.1003344 | \n",
+ "0.4947058 | \n",
+ "3.2663866 | \n",
+ "3.6014006 | \n",
+ "0.65 | \n",
+ "0.5984199 | \n",
+ "0.7166667 | \n",
+ "0.7307488 | \n",
+ "0.1638655 | \n",
+ "0.3613445 | \n",
+ "226.6386555 | \n",
+ "260.1400560 | \n",
+ "0.3258539 |
\n",
+ "| 7 | \n",
+ "0.1505017 | \n",
+ "0.3581829 | \n",
+ "2.0100840 | \n",
+ "3.0709617 | \n",
+ "0.4 | \n",
+ "0.4130805 | \n",
+ "0.6111111 | \n",
+ "0.6248593 | \n",
+ "0.1008403 | \n",
+ "0.4621849 | \n",
+ "101.0084034 | \n",
+ "207.0961718 | \n",
+ "0.3891160 |
\n",
+ "| 8 | \n",
+ "0.2006689 | \n",
+ "0.3002017 | \n",
+ "2.0100840 | \n",
+ "2.8057423 | \n",
+ "0.4 | \n",
+ "0.3274207 | \n",
+ "0.5583333 | \n",
+ "0.5504997 | \n",
+ "0.1008403 | \n",
+ "0.5630252 | \n",
+ "101.0084034 | \n",
+ "180.5742297 | \n",
+ "0.4523780 |
\n",
+ "| 9 | \n",
+ "0.3001672 | \n",
+ "0.2236892 | \n",
+ "1.1824024 | \n",
+ "2.2676435 | \n",
+ "0.2352941 | \n",
+ "0.2577997 | \n",
+ "0.4512535 | \n",
+ "0.4534766 | \n",
+ "0.1176471 | \n",
+ "0.6806723 | \n",
+ "18.2402373 | \n",
+ "126.7643548 | \n",
+ "0.4750355 |
\n",
+ "| 10 | \n",
+ "0.4005017 | \n",
+ "0.1761387 | \n",
+ "1.0469188 | \n",
+ "1.9618252 | \n",
+ "0.2083333 | \n",
+ "0.1963835 | \n",
+ "0.3903967 | \n",
+ "0.3890691 | \n",
+ "0.1050420 | \n",
+ "0.7857143 | \n",
+ "4.6918768 | \n",
+ "96.1825231 | \n",
+ "0.4809126 |
\n",
+ "| 11 | \n",
+ "0.5 | \n",
+ "0.1438673 | \n",
+ "0.5489725 | \n",
+ "1.6806723 | \n",
+ "0.1092437 | \n",
+ "0.1582719 | \n",
+ "0.3344482 | \n",
+ "0.3431412 | \n",
+ "0.0546218 | \n",
+ "0.8403361 | \n",
+ "-45.1027470 | \n",
+ "68.0672269 | \n",
+ "0.4248873 |
\n",
+ "| 12 | \n",
+ "0.6003344 | \n",
+ "0.1214381 | \n",
+ "0.5025210 | \n",
+ "1.4837668 | \n",
+ "0.1 | \n",
+ "0.1325918 | \n",
+ "0.2952646 | \n",
+ "0.3079519 | \n",
+ "0.0504202 | \n",
+ "0.8907563 | \n",
+ "-49.7478992 | \n",
+ "48.3766766 | \n",
+ "0.3625726 |
\n",
+ "| 13 | \n",
+ "0.6998328 | \n",
+ "0.0997817 | \n",
+ "0.5067439 | \n",
+ "1.3448591 | \n",
+ "0.1008403 | \n",
+ "0.1100559 | \n",
+ "0.2676225 | \n",
+ "0.2798162 | \n",
+ "0.0504202 | \n",
+ "0.9411765 | \n",
+ "-49.3256126 | \n",
+ "34.4859091 | \n",
+ "0.3013017 |
\n",
+ "| 14 | \n",
+ "0.8001672 | \n",
+ "0.0798720 | \n",
+ "0.2512605 | \n",
+ "1.2077307 | \n",
+ "0.05 | \n",
+ "0.0891178 | \n",
+ "0.2403344 | \n",
+ "0.2559041 | \n",
+ "0.0252101 | \n",
+ "0.9663866 | \n",
+ "-74.8739496 | \n",
+ "20.7730741 | \n",
+ "0.2075139 |
\n",
+ "| 15 | \n",
+ "0.8996656 | \n",
+ "0.0556184 | \n",
+ "0.2111433 | \n",
+ "1.0975134 | \n",
+ "0.0420168 | \n",
+ "0.0677374 | \n",
+ "0.2184015 | \n",
+ "0.2350939 | \n",
+ "0.0210084 | \n",
+ "0.9873950 | \n",
+ "-78.8856719 | \n",
+ "9.7513355 | \n",
+ "0.1095244 |
\n",
+ "| 16 | \n",
+ "1.0 | \n",
+ "0.0074980 | \n",
+ "0.1256303 | \n",
+ "1.0 | \n",
+ "0.025 | \n",
+ "0.0397575 | \n",
+ "0.1989967 | \n",
+ "0.2154949 | \n",
+ "0.0126050 | \n",
+ "1.0 | \n",
+ "-87.4369748 | \n",
+ "0.0 | \n",
+ "0.0 |
\n",
+ "
\n",
+ "
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ " Scoring History: \n",
+ " | \n",
+ "timestamp | \n",
+ "duration | \n",
+ "iteration | \n",
+ "lambda | \n",
+ "predictors | \n",
+ "deviance_train | \n",
+ "deviance_test | \n",
+ "alpha | \n",
+ "iterations | \n",
+ "training_rmse | \n",
+ "training_logloss | \n",
+ "training_r2 | \n",
+ "training_auc | \n",
+ "training_pr_auc | \n",
+ "training_lift | \n",
+ "training_classification_error | \n",
+ "validation_rmse | \n",
+ "validation_logloss | \n",
+ "validation_r2 | \n",
+ "validation_auc | \n",
+ "validation_pr_auc | \n",
+ "validation_lift | \n",
+ "validation_classification_error |
|---|
\n",
+ " | \n",
+ "2023-10-05 18:17:37 | \n",
+ " 0.000 sec | \n",
+ "1 | \n",
+ ".25E0 | \n",
+ "1 | \n",
+ "1.0045100 | \n",
+ "0.9980509 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:37 | \n",
+ " 0.061 sec | \n",
+ "3 | \n",
+ ".22E0 | \n",
+ "2 | \n",
+ "0.9946355 | \n",
+ "0.9864127 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:37 | \n",
+ " 0.101 sec | \n",
+ "5 | \n",
+ ".2E0 | \n",
+ "2 | \n",
+ "0.9859035 | \n",
+ "0.9760418 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:37 | \n",
+ " 0.130 sec | \n",
+ "7 | \n",
+ ".19E0 | \n",
+ "2 | \n",
+ "0.9781697 | \n",
+ "0.9667699 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:37 | \n",
+ " 0.164 sec | \n",
+ "9 | \n",
+ ".17E0 | \n",
+ "3 | \n",
+ "0.9686784 | \n",
+ "0.9544753 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:37 | \n",
+ " 0.201 sec | \n",
+ "11 | \n",
+ ".15E0 | \n",
+ "3 | \n",
+ "0.9587676 | \n",
+ "0.9411532 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:37 | \n",
+ " 0.220 sec | \n",
+ "13 | \n",
+ ".14E0 | \n",
+ "3 | \n",
+ "0.9500953 | \n",
+ "0.9293709 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:37 | \n",
+ " 0.268 sec | \n",
+ "15 | \n",
+ ".13E0 | \n",
+ "3 | \n",
+ "0.9424603 | \n",
+ "0.9188844 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:37 | \n",
+ " 0.295 sec | \n",
+ "17 | \n",
+ ".12E0 | \n",
+ "5 | \n",
+ "0.9341431 | \n",
+ "0.9080921 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:37 | \n",
+ " 0.351 sec | \n",
+ "19 | \n",
+ ".11E0 | \n",
+ "5 | \n",
+ "0.9236449 | \n",
+ "0.8955093 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ "| --- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- | \n",
+ "--- |
\n",
+ " | \n",
+ "2023-10-05 18:17:38 | \n",
+ " 1.462 sec | \n",
+ "89 | \n",
+ ".13E-2 | \n",
+ "18 | \n",
+ "0.8045920 | \n",
+ "0.7656455 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:38 | \n",
+ " 1.467 sec | \n",
+ "90 | \n",
+ ".12E-2 | \n",
+ "18 | \n",
+ "0.8044582 | \n",
+ "0.7656103 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:38 | \n",
+ " 1.472 sec | \n",
+ "91 | \n",
+ ".11E-2 | \n",
+ "18 | \n",
+ "0.8043436 | \n",
+ "0.7655884 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:38 | \n",
+ " 1.485 sec | \n",
+ "92 | \n",
+ ".1E-2 | \n",
+ "18 | \n",
+ "0.8042462 | \n",
+ "0.7655764 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:38 | \n",
+ " 1.492 sec | \n",
+ "93 | \n",
+ ".92E-3 | \n",
+ "18 | \n",
+ "0.8041639 | \n",
+ "0.7655709 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:38 | \n",
+ " 1.498 sec | \n",
+ "94 | \n",
+ ".84E-3 | \n",
+ "18 | \n",
+ "0.8040939 | \n",
+ "0.7655721 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:38 | \n",
+ " 1.502 sec | \n",
+ "95 | \n",
+ ".77E-3 | \n",
+ "18 | \n",
+ "0.8040347 | \n",
+ "0.7655776 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:38 | \n",
+ " 1.506 sec | \n",
+ "96 | \n",
+ ".7E-3 | \n",
+ "18 | \n",
+ "0.8039850 | \n",
+ "0.7655875 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:38 | \n",
+ " 1.518 sec | \n",
+ "97 | \n",
+ ".64E-3 | \n",
+ "18 | \n",
+ "0.8039428 | \n",
+ "0.7655990 | \n",
+ "0.5 | \n",
+ "None | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " | \n",
+ " |
\n",
+ " | \n",
+ "2023-10-05 18:17:38 | \n",
+ " 1.528 sec | \n",
+ "98 | \n",
+ ".58E-3 | \n",
+ "18 | \n",
+ "0.8039072 | \n",
+ "0.7656134 | \n",
+ "0.5 | \n",
+ "98 | \n",
+ "0.3522849 | \n",
+ "0.4020820 | \n",
+ "0.2282168 | \n",
+ "0.7849415 | \n",
+ "0.5691249 | \n",
+ "4.9667129 | \n",
+ "0.2032952 | \n",
+ "0.3432682 | \n",
+ "0.3827854 | \n",
+ "0.2607574 | \n",
+ "0.8173892 | \n",
+ "0.5982393 | \n",
+ "5.0252101 | \n",
+ "0.1688963 |
\n",
+ "
\n",
+ "
\n",
+ "
[66 rows x 24 columns]
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ " Variable Importances: \n",
+ " | variable | \n",
+ "relative_importance | \n",
+ "scaled_importance | \n",
+ "percentage |
|---|
\n",
+ " | DELINQ | \n",
+ "0.7970053 | \n",
+ "1.0 | \n",
+ "0.1473004 |
\n",
+ "| JOB.Sales | \n",
+ "0.7478484 | \n",
+ "0.9383230 | \n",
+ "0.1382154 |
\n",
+ "| JOB.Office | \n",
+ "0.5109144 | \n",
+ "0.6410426 | \n",
+ "0.0944259 |
\n",
+ "| JOB.Self | \n",
+ "0.4571329 | \n",
+ "0.5735632 | \n",
+ "0.0844861 |
\n",
+ "| CLAGE | \n",
+ "0.4484834 | \n",
+ "0.5627107 | \n",
+ "0.0828875 |
\n",
+ "| DEBTINC | \n",
+ "0.4467476 | \n",
+ "0.5605328 | \n",
+ "0.0825667 |
\n",
+ "| DEROG | \n",
+ "0.4263194 | \n",
+ "0.5349016 | \n",
+ "0.0787912 |
\n",
+ "| NINQ | \n",
+ "0.3037309 | \n",
+ "0.3810902 | \n",
+ "0.0561347 |
\n",
+ "| VALUE | \n",
+ "0.2412600 | \n",
+ "0.3027081 | \n",
+ "0.0445890 |
\n",
+ "| MORTDUE | \n",
+ "0.2384487 | \n",
+ "0.2991808 | \n",
+ "0.0440695 |
\n",
+ "| LOAN | \n",
+ "0.2180768 | \n",
+ "0.2736203 | \n",
+ "0.0403044 |
\n",
+ "| REASON.HomeImp | \n",
+ "0.1797686 | \n",
+ "0.2255551 | \n",
+ "0.0332244 |
\n",
+ "| CLNO | \n",
+ "0.1485205 | \n",
+ "0.1863481 | \n",
+ "0.0274492 |
\n",
+ "| REASON.DebtCon | \n",
+ "0.1373842 | \n",
+ "0.1723755 | \n",
+ "0.0253910 |
\n",
+ "| YOJ | \n",
+ "0.0438345 | \n",
+ "0.0549990 | \n",
+ "0.0081014 |
\n",
+ "| JOB.Other | \n",
+ "0.0353093 | \n",
+ "0.0443025 | \n",
+ "0.0065258 |
\n",
+ "| JOB.Mgr | \n",
+ "0.0299615 | \n",
+ "0.0375926 | \n",
+ "0.0055374 |
\n",
+ "| JOB.ProfExe | \n",
+ "0.0 | \n",
+ "0.0 | \n",
+ "0.0 |
\n",
+ "
\n",
+ "
\n",
+ "
\n",
+ "\n",
+ "[tips]\n",
+ "Use `model.explain()` to inspect the model.\n",
+ "--\n",
+ "Use `h2o.display.toggle_user_tips()` to switch on/off this section.
"
+ ],
+ "text/plain": [
+ "Model Details\n",
+ "=============\n",
+ "H2OGeneralizedLinearEstimator : Generalized Linear Modeling\n",
+ "Model Key: glmfit\n",
+ "\n",
+ "\n",
+ "GLM Model: summary\n",
+ " family link regularization lambda_search number_of_predictors_total number_of_active_predictors number_of_iterations training_frame\n",
+ "-- -------- ------ --------------------------------------------- ---------------------------------------------------------------------------- ---------------------------- ----------------------------- ---------------------- ----------------\n",
+ " binomial logit Elastic Net (alpha = 0.5, lambda = 9.244E-4 ) nlambda = 100, lambda.max = 0.2455, lambda.min = 9.244E-4, lambda.1se = -1.0 18 17 93 py_3_sid_a269\n",
+ "\n",
+ "ModelMetricsBinomialGLM: glm\n",
+ "** Reported on train data. **\n",
+ "\n",
+ "MSE: 0.12410463474559494\n",
+ "RMSE: 0.3522848772592927\n",
+ "LogLoss: 0.40208197287665776\n",
+ "AUC: 0.7849415147958837\n",
+ "AUCPR: 0.5691248967099318\n",
+ "Gini: 0.5698830295917674\n",
+ "Null degrees of freedom: 3580\n",
+ "Residual degrees of freedom: 3563\n",
+ "Null deviance: 3597.150438148379\n",
+ "Residual deviance: 2879.711089742623\n",
+ "AIC: 2915.711089742623\n",
+ "\n",
+ "Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.2574171649920362\n",
+ " 0 1 Error Rate\n",
+ "----- ---- --- ------- --------------\n",
+ "0 2446 414 0.1448 (414.0/2860.0)\n",
+ "1 314 407 0.4355 (314.0/721.0)\n",
+ "Total 2760 821 0.2033 (728.0/3581.0)\n",
+ "\n",
+ "Maximum Metrics: Maximum metrics at their respective thresholds\n",
+ "metric threshold value idx\n",
+ "--------------------------- ----------- -------- -----\n",
+ "max f1 0.257417 0.527886 207\n",
+ "max f2 0.152297 0.626112 279\n",
+ "max f0point5 0.388606 0.557126 144\n",
+ "max accuracy 0.567259 0.835242 90\n",
+ "max precision 0.998809 1 0\n",
+ "max recall 0.00123775 1 399\n",
+ "max specificity 0.998809 1 0\n",
+ "max absolute_mcc 0.266185 0.402098 203\n",
+ "max min_per_class_accuracy 0.17828 0.706643 258\n",
+ "max mean_per_class_accuracy 0.214315 0.719503 234\n",
+ "max tns 0.998809 2860 0\n",
+ "max fns 0.998809 717 0\n",
+ "max fps 0.00626272 2860 398\n",
+ "max tps 0.00123775 721 399\n",
+ "max tnr 0.998809 1 0\n",
+ "max fnr 0.998809 0.994452 0\n",
+ "max fpr 0.00626272 1 398\n",
+ "max tpr 0.00123775 1 399\n",
+ "\n",
+ "Gains/Lift Table: Avg response rate: 20.13 %, avg score: 20.13 %\n",
+ "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 kolmogorov_smirnov\n",
+ "------- -------------------------- ----------------- -------- ----------------- --------------- --------- -------------------------- ------------------ -------------- ------------------------- -------- ----------------- --------------------\n",
+ "1 0.0100531 0.946682 4.96671 4.96671 1 0.980507 1 0.980507 0.0499307 0.0499307 396.671 396.671 0.0499307\n",
+ "2 0.0201061 0.85234 4.82875 4.89773 0.972222 0.898794 0.986111 0.93965 0.0485437 0.0984743 382.875 389.773 0.0981247\n",
+ "3 0.0301592 0.747956 3.863 4.55282 0.777778 0.798407 0.916667 0.892569 0.038835 0.137309 286.3 355.282 0.134162\n",
+ "4 0.0402122 0.691234 3.17318 4.20791 0.638889 0.715245 0.847222 0.848238 0.0319001 0.169209 217.318 320.791 0.161517\n",
+ "5 0.0502653 0.62673 3.58707 4.08374 0.722222 0.655894 0.822222 0.809769 0.036061 0.20527 258.707 308.374 0.194082\n",
+ "6 0.100251 0.430246 2.44174 3.26503 0.49162 0.519975 0.657382 0.665276 0.122053 0.327323 144.174 226.503 0.284316\n",
+ "7 0.150237 0.345918 2.05328 2.86186 0.413408 0.383856 0.576208 0.571644 0.102635 0.429958 105.328 186.186 0.350238\n",
+ "8 0.200223 0.284421 1.72031 2.57687 0.346369 0.311903 0.518828 0.506799 0.0859917 0.51595 72.0314 157.687 0.395321\n",
+ "9 0.300195 0.209844 1.33186 2.16225 0.268156 0.243115 0.435349 0.418986 0.133148 0.649098 33.1856 116.225 0.436861\n",
+ "10 0.400168 0.169573 0.887904 1.84389 0.178771 0.188252 0.371249 0.361343 0.0887656 0.737864 -11.2096 84.3888 0.422829\n",
+ "11 0.50014 0.138032 0.679801 1.6112 0.136872 0.152531 0.3244 0.319604 0.0679612 0.805825 -32.0199 61.1201 0.382748\n",
+ "12 0.600112 0.114623 0.541066 1.43293 0.108939 0.125884 0.288506 0.287332 0.0540915 0.859917 -45.8934 43.2928 0.325301\n",
+ "13 0.700084 0.0945753 0.527193 1.30359 0.106145 0.104779 0.262465 0.261263 0.0527046 0.912621 -47.2807 30.3589 0.266118\n",
+ "14 0.800056 0.0736148 0.388458 1.18924 0.0782123 0.0841132 0.239442 0.239127 0.038835 0.951456 -61.1542 18.9237 0.189568\n",
+ "15 0.900028 0.0521793 0.235849 1.08334 0.047486 0.0627694 0.21812 0.219538 0.0235784 0.975035 -76.4151 8.33382 0.0939158\n",
+ "16 1 0.00111668 0.249723 1 0.0502793 0.0375156 0.20134 0.201341 0.0249653 1 -75.0277 0 0\n",
+ "\n",
+ "ModelMetricsBinomialGLM: glm\n",
+ "** Reported on validation data. **\n",
+ "\n",
+ "MSE: 0.1178330502946334\n",
+ "RMSE: 0.3432681900418875\n",
+ "LogLoss: 0.3827854417227382\n",
+ "AUC: 0.8173891686110769\n",
+ "AUCPR: 0.5982392824442148\n",
+ "Gini: 0.6347783372221538\n",
+ "Null degrees of freedom: 1195\n",
+ "Residual degrees of freedom: 1178\n",
+ "Null deviance: 1193.6689319811862\n",
+ "Residual deviance: 915.6227766007898\n",
+ "AIC: 951.6227766007898\n",
+ "\n",
+ "Confusion Matrix (Act/Pred) for max f1 @ threshold = 0.3129932533175332\n",
+ " 0 1 Error Rate\n",
+ "----- --- --- ------- --------------\n",
+ "0 863 95 0.0992 (95.0/958.0)\n",
+ "1 107 131 0.4496 (107.0/238.0)\n",
+ "Total 970 226 0.1689 (202.0/1196.0)\n",
+ "\n",
+ "Maximum Metrics: Maximum metrics at their respective thresholds\n",
+ "metric threshold value idx\n",
+ "--------------------------- ----------- -------- -----\n",
+ "max f1 0.312993 0.564655 155\n",
+ "max f2 0.187746 0.655148 231\n",
+ "max f0point5 0.429613 0.60794 107\n",
+ "max accuracy 0.443849 0.846154 105\n",
+ "max precision 0.993986 1 0\n",
+ "max recall 0.00965813 1 398\n",
+ "max specificity 0.993986 1 0\n",
+ "max absolute_mcc 0.312993 0.460207 155\n",
+ "max min_per_class_accuracy 0.195405 0.747899 225\n",
+ "max mean_per_class_accuracy 0.191454 0.750803 228\n",
+ "max tns 0.993986 958 0\n",
+ "max fns 0.993986 237 0\n",
+ "max fps 0.00749797 958 399\n",
+ "max tps 0.00965813 238 398\n",
+ "max tnr 0.993986 1 0\n",
+ "max fnr 0.993986 0.995798 0\n",
+ "max fpr 0.00749797 1 399\n",
+ "max tpr 0.00965813 1 398\n",
+ "\n",
+ "Gains/Lift Table: Avg response rate: 19.90 %, avg score: 21.55 %\n",
+ "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 kolmogorov_smirnov\n",
+ "------- -------------------------- ----------------- -------- ----------------- --------------- --------- -------------------------- ------------------ -------------- ------------------------- -------- ----------------- --------------------\n",
+ "1 0.0100334 0.942283 5.02521 5.02521 1 0.969816 1 0.969816 0.0504202 0.0504202 402.521 402.521 0.0504202\n",
+ "2 0.0200669 0.885942 4.60644 4.81583 0.916667 0.913326 0.958333 0.941571 0.0462185 0.0966387 360.644 381.583 0.0955948\n",
+ "3 0.0301003 0.833388 3.76891 4.46685 0.75 0.866251 0.888889 0.916464 0.0378151 0.134454 276.891 346.685 0.130278\n",
+ "4 0.0401338 0.784858 3.76891 4.29237 0.75 0.815402 0.854167 0.891199 0.0378151 0.172269 276.891 329.237 0.164962\n",
+ "5 0.0501672 0.721196 2.51261 3.93641 0.5 0.750593 0.783333 0.863078 0.0252101 0.197479 151.261 293.641 0.183909\n",
+ "6 0.100334 0.494706 3.26639 3.6014 0.65 0.59842 0.716667 0.730749 0.163866 0.361345 226.639 260.14 0.325854\n",
+ "7 0.150502 0.358183 2.01008 3.07096 0.4 0.41308 0.611111 0.624859 0.10084 0.462185 101.008 207.096 0.389116\n",
+ "8 0.200669 0.300202 2.01008 2.80574 0.4 0.327421 0.558333 0.5505 0.10084 0.563025 101.008 180.574 0.452378\n",
+ "9 0.300167 0.223689 1.1824 2.26764 0.235294 0.2578 0.451253 0.453477 0.117647 0.680672 18.2402 126.764 0.475036\n",
+ "10 0.400502 0.176139 1.04692 1.96183 0.208333 0.196383 0.390397 0.389069 0.105042 0.785714 4.69188 96.1825 0.480913\n",
+ "11 0.5 0.143867 0.548973 1.68067 0.109244 0.158272 0.334448 0.343141 0.0546218 0.840336 -45.1027 68.0672 0.424887\n",
+ "12 0.600334 0.121438 0.502521 1.48377 0.1 0.132592 0.295265 0.307952 0.0504202 0.890756 -49.7479 48.3767 0.362573\n",
+ "13 0.699833 0.0997817 0.506744 1.34486 0.10084 0.110056 0.267622 0.279816 0.0504202 0.941176 -49.3256 34.4859 0.301302\n",
+ "14 0.800167 0.079872 0.251261 1.20773 0.05 0.0891178 0.240334 0.255904 0.0252101 0.966387 -74.8739 20.7731 0.207514\n",
+ "15 0.899666 0.0556184 0.211143 1.09751 0.0420168 0.0677374 0.218401 0.235094 0.0210084 0.987395 -78.8857 9.75134 0.109524\n",
+ "16 1 0.00749797 0.12563 1 0.025 0.0397575 0.198997 0.215495 0.012605 1 -87.437 0 0\n",
+ "\n",
+ "Scoring History: \n",
+ " timestamp duration iteration lambda predictors deviance_train deviance_test alpha iterations training_rmse training_logloss training_r2 training_auc training_pr_auc training_lift training_classification_error validation_rmse validation_logloss validation_r2 validation_auc validation_pr_auc validation_lift validation_classification_error\n",
+ "--- ------------------- ---------- ----------- -------- ------------ ------------------ ------------------ ------- ------------ ------------------ ------------------- ------------------- ------------------ ------------------ ----------------- ------------------------------- ------------------ -------------------- ------------------- ------------------ ------------------- ----------------- ---------------------------------\n",
+ " 2023-10-05 18:17:37 0.000 sec 1 .25E0 1 1.0045100357856405 0.9980509464725638 0.5\n",
+ " 2023-10-05 18:17:37 0.061 sec 3 .22E0 2 0.9946354860817245 0.9864126793254803 0.5\n",
+ " 2023-10-05 18:17:37 0.101 sec 5 .2E0 2 0.9859035403430406 0.9760418435888577 0.5\n",
+ " 2023-10-05 18:17:37 0.130 sec 7 .19E0 2 0.9781697250290142 0.9667699237562071 0.5\n",
+ " 2023-10-05 18:17:37 0.164 sec 9 .17E0 3 0.9686784218176835 0.9544752889432686 0.5\n",
+ " 2023-10-05 18:17:37 0.201 sec 11 .15E0 3 0.9587676036112146 0.9411531794664825 0.5\n",
+ " 2023-10-05 18:17:37 0.220 sec 13 .14E0 3 0.9500952786201636 0.9293708568093554 0.5\n",
+ " 2023-10-05 18:17:37 0.268 sec 15 .13E0 3 0.9424603199349196 0.918884378267228 0.5\n",
+ " 2023-10-05 18:17:37 0.295 sec 17 .12E0 5 0.934143096516142 0.9080921324097252 0.5\n",
+ " 2023-10-05 18:17:37 0.351 sec 19 .11E0 5 0.9236448771247918 0.8955093341627816 0.5\n",
+ "--- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- ---\n",
+ " 2023-10-05 18:17:38 1.462 sec 89 .13E-2 18 0.8045920176677348 0.7656454925772931 0.5\n",
+ " 2023-10-05 18:17:38 1.467 sec 90 .12E-2 18 0.8044581526528738 0.7656102932355412 0.5\n",
+ " 2023-10-05 18:17:38 1.472 sec 91 .11E-2 18 0.804343644390629 0.7655884396559107 0.5\n",
+ " 2023-10-05 18:17:38 1.485 sec 92 .1E-2 18 0.804246194799186 0.7655763506822736 0.5\n",
+ " 2023-10-05 18:17:38 1.492 sec 93 .92E-3 18 0.8041639457533155 0.7655708834454764 0.5\n",
+ " 2023-10-05 18:17:38 1.498 sec 94 .84E-3 18 0.8040939470898276 0.7655720881120318 0.5\n",
+ " 2023-10-05 18:17:38 1.502 sec 95 .77E-3 18 0.8040347348308917 0.7655776424675973 0.5\n",
+ " 2023-10-05 18:17:38 1.506 sec 96 .7E-3 18 0.8039850013216244 0.7655874603666687 0.5\n",
+ " 2023-10-05 18:17:38 1.518 sec 97 .64E-3 18 0.8039427775722173 0.7655990044572286 0.5\n",
+ " 2023-10-05 18:17:38 1.528 sec 98 .58E-3 18 0.8039071637538451 0.7656134388669195 0.5 98 0.3522848772592927 0.40208197287665776 0.22821675700859534 0.7849415147958837 0.5691248967099318 4.966712898751734 0.20329516894722144 0.3432681900418875 0.3827854417227382 0.26075735307166414 0.8173891686110769 0.5982392824442148 5.025210084033613 0.1688963210702341\n",
+ "[66 rows x 24 columns]\n",
+ "\n",
+ "\n",
+ "Variable Importances: \n",
+ "variable relative_importance scaled_importance percentage\n",
+ "-------------- --------------------- ------------------- ------------\n",
+ "DELINQ 0.797005 1 0.1473\n",
+ "JOB.Sales 0.747848 0.938323 0.138215\n",
+ "JOB.Office 0.510914 0.641043 0.0944259\n",
+ "JOB.Self 0.457133 0.573563 0.0844861\n",
+ "CLAGE 0.448483 0.562711 0.0828875\n",
+ "DEBTINC 0.446748 0.560533 0.0825667\n",
+ "DEROG 0.426319 0.534902 0.0787912\n",
+ "NINQ 0.303731 0.38109 0.0561347\n",
+ "VALUE 0.24126 0.302708 0.044589\n",
+ "MORTDUE 0.238449 0.299181 0.0440695\n",
+ "LOAN 0.218077 0.27362 0.0403044\n",
+ "REASON.HomeImp 0.179769 0.225555 0.0332244\n",
+ "CLNO 0.14852 0.186348 0.0274492\n",
+ "REASON.DebtCon 0.137384 0.172376 0.025391\n",
+ "YOJ 0.0438345 0.054999 0.00810138\n",
+ "JOB.Other 0.0353093 0.0443025 0.00652578\n",
+ "JOB.Mgr 0.0299615 0.0375926 0.0055374\n",
+ "JOB.ProfExe 0 0 0\n",
+ "\n",
+ "[tips]\n",
+ "Use `model.explain()` to inspect the model.\n",
+ "--\n",
+ "Use `h2o.display.toggle_user_tips()` to switch on/off this section."
+ ]
},
"execution_count": 7,
"metadata": {},
@@ -243,7 +2267,7 @@
"name": "stdout",
"output_type": "stream",
"text": [
- "[[0.551348008992684, 0.8486897717666948]]\n"
+ "[[0.5513512979207219, 0.8486897717666948]]\n"
]
}
],
@@ -253,6 +2277,15 @@
"print(glm_performance.accuracy())"
]
},
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "glm.pre"
+ ]
+ },
{
"cell_type": "markdown",
"metadata": {
@@ -321,14 +2354,14 @@
"name": "stdout",
"output_type": "stream",
"text": [
- "inputVar.json was successfully written and saved to ~\\examples\\data\\hmeqModels\\H2OBinaryGLM\\inputVar.json\n",
- "inputVar.json was successfully written and saved to ~\\examples\\data\\hmeqModels\\H2OMOJOGLM\\inputVar.json\n",
- "outputVar.json was successfully written and saved to ~\\examples\\data\\hmeqModels\\H2OBinaryGLM\\outputVar.json\n",
- "outputVar.json was successfully written and saved to ~\\examples\\data\\hmeqModels\\H2OMOJOGLM\\outputVar.json\n",
- "ModelProperties.json was successfully written and saved to ~\\examples\\data\\hmeqModels\\H2OBinaryGLM\\ModelProperties.json\n",
- "ModelProperties.json was successfully written and saved to ~\\examples\\data\\hmeqModels\\H2OMOJOGLM\\ModelProperties.json\n",
- "fileMetadata.json was successfully written and saved to ~\\examples\\data\\hmeqModels\\H2OBinaryGLM\\fileMetadata.json\n",
- "fileMetadata.json was successfully written and saved to ~\\examples\\data\\hmeqModels\\H2OMOJOGLM\\fileMetadata.json\n"
+ "inputVar.json was successfully written and saved to /Users/dalmoo/Documents/GitHub/python-sasctl/examples/data/hmeqModels/H2OBinaryGLM/inputVar.json\n",
+ "inputVar.json was successfully written and saved to /Users/dalmoo/Documents/GitHub/python-sasctl/examples/data/hmeqModels/H2OMOJOGLM/inputVar.json\n",
+ "outputVar.json was successfully written and saved to /Users/dalmoo/Documents/GitHub/python-sasctl/examples/data/hmeqModels/H2OBinaryGLM/outputVar.json\n",
+ "outputVar.json was successfully written and saved to /Users/dalmoo/Documents/GitHub/python-sasctl/examples/data/hmeqModels/H2OMOJOGLM/outputVar.json\n",
+ "ModelProperties.json was successfully written and saved to /Users/dalmoo/Documents/GitHub/python-sasctl/examples/data/hmeqModels/H2OBinaryGLM/ModelProperties.json\n",
+ "ModelProperties.json was successfully written and saved to /Users/dalmoo/Documents/GitHub/python-sasctl/examples/data/hmeqModels/H2OMOJOGLM/ModelProperties.json\n",
+ "fileMetadata.json was successfully written and saved to /Users/dalmoo/Documents/GitHub/python-sasctl/examples/data/hmeqModels/H2OBinaryGLM/fileMetadata.json\n",
+ "fileMetadata.json was successfully written and saved to /Users/dalmoo/Documents/GitHub/python-sasctl/examples/data/hmeqModels/H2OMOJOGLM/fileMetadata.json\n"
]
}
],
@@ -351,7 +2384,7 @@
" model_name=model_prefix + \"_binary\",\n",
" model_desc=\"Binary H2O model.\",\n",
" target_variable=y,\n",
- " target_values=[\"1\", \"0\"],\n",
+ " target_values=[\"0\", \"1\"],\n",
" json_path=binary_folder,\n",
" modeler=\"sasdemo\"\n",
")\n",
@@ -359,7 +2392,7 @@
" model_name=model_prefix + \"_mojo\",\n",
" model_desc=\"MOJO H2O model.\",\n",
" target_variable=y,\n",
- " target_values=[\"1\", \"0\"],\n",
+ " target_values=[\"0\", \"1\"],\n",
" json_path=mojo_folder,\n",
" modeler=\"sasdemo\"\n",
")\n",
@@ -435,36 +2468,45 @@
},
{
"cell_type": "markdown",
- "source": [
- "### Run a Score Test in SAS Model Manager"
- ],
"metadata": {
"collapsed": false
- }
+ },
+ "source": [
+ "### Run a Score Test in SAS Model Manager"
+ ]
},
{
"cell_type": "code",
"execution_count": 15,
- "outputs": [],
- "source": [
- "# Publish the model to the SAS Microanalytic Score destination in SAS Model Manager\n",
- "module = publish_model(mojo_model[0], \"maslocal\", name=\"HMEQMOJO_publish\", replace=True)"
- ],
"metadata": {
- "collapsed": false,
"ExecuteTime": {
"end_time": "2023-06-14T17:18:16.283114900Z",
"start_time": "2023-06-14T17:18:08.890088300Z"
- }
- }
+ },
+ "collapsed": false
+ },
+ "outputs": [],
+ "source": [
+ "# Publish the model to the SAS Microanalytic Score destination in SAS Model Manager\n",
+ "module = publish_model(mojo_model[0], \"maslocal\", name=\"HMEQMOJO_publish\", replace=True)"
+ ]
},
{
"cell_type": "code",
"execution_count": 16,
+ "metadata": {
+ "ExecuteTime": {
+ "end_time": "2023-06-14T17:18:34.236925200Z",
+ "start_time": "2023-06-14T17:18:34.209029800Z"
+ },
+ "collapsed": false
+ },
"outputs": [
{
"data": {
- "text/plain": ""
+ "text/plain": [
+ ""
+ ]
},
"execution_count": 16,
"metadata": {},
@@ -474,18 +2516,18 @@
"source": [
"# Instantiate a API call logger to visualize score calls in realtime\n",
"sess.add_stderr_logger(level=20)"
- ],
- "metadata": {
- "collapsed": false,
- "ExecuteTime": {
- "end_time": "2023-06-14T17:18:34.236925200Z",
- "start_time": "2023-06-14T17:18:34.209029800Z"
- }
- }
+ ]
},
{
"cell_type": "code",
"execution_count": 17,
+ "metadata": {
+ "ExecuteTime": {
+ "end_time": "2023-06-14T17:18:47.064393600Z",
+ "start_time": "2023-06-14T17:18:37.850162200Z"
+ },
+ "collapsed": false
+ },
"outputs": [
{
"name": "stderr",
@@ -521,18 +2563,18 @@
"# Step through the rows of data and collect the score from SAS Microanalytic Score publish destination\n",
"for index, row in X.iterrows():\n",
" result.append(module.score(row))"
- ],
- "metadata": {
- "collapsed": false,
- "ExecuteTime": {
- "end_time": "2023-06-14T17:18:47.064393600Z",
- "start_time": "2023-06-14T17:18:37.850162200Z"
- }
- }
+ ]
},
{
"cell_type": "code",
"execution_count": 18,
+ "metadata": {
+ "ExecuteTime": {
+ "end_time": "2023-06-14T17:19:21.811834300Z",
+ "start_time": "2023-06-14T17:19:21.772752700Z"
+ },
+ "collapsed": false
+ },
"outputs": [
{
"name": "stdout",
@@ -554,33 +2596,23 @@
"source": [
"# Scoring results\n",
"pp.pprint(result)"
- ],
- "metadata": {
- "collapsed": false,
- "ExecuteTime": {
- "end_time": "2023-06-14T17:19:21.811834300Z",
- "start_time": "2023-06-14T17:19:21.772752700Z"
- }
- }
+ ]
},
{
"cell_type": "code",
"execution_count": null,
- "outputs": [],
- "source": [],
"metadata": {
"collapsed": false
- }
+ },
+ "outputs": [],
+ "source": []
}
],
"metadata": {
- "interpreter": {
- "hash": "f9708d3f38eeab835578f0695c8890716ee809285281a28db6e379a5abca1310"
- },
"kernelspec": {
- "display_name": "dev-py38",
+ "display_name": "yeehaw",
"language": "python",
- "name": "dev-py38"
+ "name": "python3"
},
"language_info": {
"codemirror_mode": {
diff --git a/examples/pzmm_tensorflow_keras_model_import.ipynb b/examples/pzmm_tensorflow_keras_model_import.ipynb
index df0fa399..e45e702b 100644
--- a/examples/pzmm_tensorflow_keras_model_import.ipynb
+++ b/examples/pzmm_tensorflow_keras_model_import.ipynb
@@ -753,7 +753,7 @@
" input_data=x, # What does example input data look like?\n",
" predict_method=[model.predict, [int, int]], # What is the predict method and what does it return?\n",
" overwrite_model=True, # Overwrite model if it arleady exists\n",
- " target_values=[\"1\", \"0\"], # What are the expecte values of the target variable?\n",
+ " target_values=[\"0\", \"1\"], # What are the expecte values of the target variable?\n",
" score_metrics=score_metrics, # What are the output variables?\n",
" model_file_name = model_prefix + \".h5\", # How was the model file serialized?\n",
" missing_values = True, # Does the data include missing values?\n",
diff --git a/src/sasctl/pzmm/import_model.py b/src/sasctl/pzmm/import_model.py
index 899042ce..dc66deab 100644
--- a/src/sasctl/pzmm/import_model.py
+++ b/src/sasctl/pzmm/import_model.py
@@ -198,6 +198,7 @@ def import_model(
predict_threshold: Optional[float] = None,
target_values: Optional[List[str]] = None,
overwrite_project_properties: Optional[bool] = False,
+ target_index: Optional[int] = None,
**kwargs,
) -> Tuple[RestObj, Union[dict, str, Path]]:
"""
@@ -275,10 +276,16 @@ def import_model(
target_values : list of strings, optional
A list of target values for the target variable. This argument and the
score_metrics argument dictate the handling of the predicted values from
- the prediction method. The default value is None.
+ the prediction method. The order of the target values should reflect the
+ order of the related probabilities in the model. The default value is None.
overwrite_project_properties : bool, optional
Set whether the project properties should be overwritten when attempting to
import the model. The default value is False.
+ target_index : int, optional
+ Sets the index of success for a binary model. If target_values are given, this
+ index should match the index of the target outcome in target_values. If target_values
+ are not given, this index should indicate whether the the target probability variable
+ is the first or second variable returned by the model. The default value is 1.
kwargs : dict, optional
Other keyword arguments are passed to the following function:
* sasctl.pzmm.ScoreCode.write_score_code(...,
@@ -352,6 +359,7 @@ def import_model(
target_values=target_values,
missing_values=missing_values,
score_cas=score_cas,
+ target_index=target_index,
**kwargs,
)
if score_code_dict:
@@ -451,6 +459,7 @@ def import_model(
target_values=target_values,
missing_values=missing_values,
score_cas=score_cas,
+ target_index=target_index,
**kwargs,
)
if score_code_dict:
diff --git a/src/sasctl/pzmm/write_score_code.py b/src/sasctl/pzmm/write_score_code.py
index 7af3c3f5..056097c6 100644
--- a/src/sasctl/pzmm/write_score_code.py
+++ b/src/sasctl/pzmm/write_score_code.py
@@ -35,6 +35,7 @@ def write_score_code(
missing_values: Union[bool, list, DataFrame] = False,
score_cas: Optional[bool] = True,
score_code_path: Union[Path, str, None] = None,
+ target_index: Optional[int] = None,
**kwargs,
) -> Union[dict, None]:
"""
@@ -129,6 +130,11 @@ def write_score_code(
score_code_path : str or Path, optional
Path for output score code file(s) to be generated. If no value is supplied
a dict is returned instead. The default value is None.
+ target_index : int, optional
+ Sets the index of success for a binary model. If target_values are given, this
+ index should match the index of the target outcome in target_values. If target_values
+ are not given, this index should indicate whether the the target probability variable
+ is the first or second variable returned by the model. The default value is 1.
kwargs
Other keyword arguments are passed to one of the following functions:
* sasctl.pzmm.ScoreCode._write_imports(pickle_type, mojo_model=None,
@@ -245,6 +251,7 @@ def score(var1, var2, var3, var4):
predict_method[1],
target_values=target_values,
predict_threshold=predict_threshold,
+ target_index=target_index,
h2o_model=True,
)
else:
@@ -275,6 +282,7 @@ def score(var1, var2, var3, var4):
predict_method[1],
target_values=target_values,
predict_threshold=predict_threshold,
+ target_index=target_index,
)
if missing_values:
@@ -1082,6 +1090,7 @@ def _predictions_to_metrics(
target_values: Optional[List[str]] = None,
predict_threshold: Optional[float] = None,
h2o_model: Optional[bool] = False,
+ target_index: Optional[int] = 1,
) -> None:
"""
Using the provided arguments, write in to the score code the method for handling
@@ -1106,6 +1115,11 @@ def _predictions_to_metrics(
h2o_model : bool, optional
Flag to indicate that the model is an H2O.ai model. The default value is
False.
+ target_index : int, optional
+ Sets the index of success for a binary model. If target_values are given, this
+ index should match the index of the target outcome in target_values. If target_values
+ are not given, this index should indicate whether the the target probability variable
+ is the first or second variable returned by the model. The default value is 1.
"""
if len(metrics) == 1 and isinstance(metrics, list):
# Flatten single valued list
@@ -1122,7 +1136,12 @@ def _predictions_to_metrics(
# Binary classification model
elif len(target_values) == 2:
cls._binary_target(
- metrics, target_values, predict_returns, predict_threshold, h2o_model
+ metrics,
+ target_values,
+ predict_returns,
+ predict_threshold,
+ target_index,
+ h2o_model,
)
# Multiclass classification model
elif len(target_values) > 2:
@@ -1173,27 +1192,27 @@ def _no_targets_no_thresholds(
)
"""
if input_array.shape[0] == 1:
- Classification = prediction[0]
+ Classification = prediction[1][0]
return Classification
else:
- output_table = pd.DataFrame({'Classification': prediction})
+ output_table = prediction.drop(prediction.columns[1:], axis=1)
+ output_table.columns = ['Classification']
return output_table
"""
else:
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}{metrics} = prediction[0]\n{'':8}return {metrics}\n"
+ f"{'':8}{metrics} = prediction[0][0]\n{'':8}return {metrics}\n"
f"{'':4}else:\n"
f"{'':8}output_table = pd.DataFrame({{'{metrics}': prediction}})\n"
f"{'':8}return output_table"
)
"""
if input_array.shape[0] == 1:
- Classification = prediction[1][0]
+ Classification = prediction[0][0]
return Classification
else:
- output_table = prediction.drop(prediction.columns[1:], axis=1)
- output_table.columns = ['Classification']
+ output_table = pd.DataFrame({'Classification': prediction})
return output_table
"""
else:
@@ -1228,7 +1247,7 @@ def _no_targets_no_thresholds(
else:
cls.score_code += f"{'':4}if input_array.shape[0] == 1:\n"
for i in range(len(metrics)):
- cls.score_code += f"{'':8}{metrics[i]} = prediction[{i}]\n"
+ cls.score_code += f"{'':8}{metrics[i]} = prediction[0][{i}]\n"
cls.score_code += f"\n{'':8}return {', '.join(metrics)}\n"
cls.score_code += (
f"{'':4}else:\n"
@@ -1237,10 +1256,10 @@ def _no_targets_no_thresholds(
)
"""
if input_array.shape[0] == 1:
- Classification = prediction[0]
- Proba_A = prediction[1]
- Proba_B = prediction[2]
- Proba_C = prediction[3]
+ Classification = prediction[0][0]
+ Proba_A = prediction[0][1]
+ Proba_B = prediction[0][2]
+ Proba_C = prediction[0][3]
return Classification, Proba_A, Proba_B, Proba_C
else:
@@ -1256,6 +1275,7 @@ def _binary_target(
target_values: List[str],
returns: List[Any],
threshold: Optional[float] = None,
+ target_index: Optional[int] = 1,
h2o_model: Optional[bool] = None,
) -> None:
"""
@@ -1276,6 +1296,9 @@ def _binary_target(
h2o_model : bool, optional
Flag to indicate that the model is an H2O.ai model. The default value is
False.
+ target_index : int, optional
+ Sets the index of the probability value to be returned from a binary model. The
+ default value is two for h2o models, and one otherwise.
"""
if not threshold:
# Set default threshold
@@ -1299,14 +1322,13 @@ def _binary_target(
if h2o_model:
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}if prediction[1][2] > {threshold}:\n"
- f"{'':12}{metrics} = \"{target_values[0]}\"\n"
+ f"{'':8}if prediction[1][{target_index+1}] > {threshold}:\n"
+ f"{'':12}{metrics} = \"{target_values[target_index]}\"\n"
f"{'':8}else:\n"
- f"{'':12}{metrics} = \"{target_values[1]}\"\n"
+ f"{'':12}{metrics} = \"{target_values[abs(target_index-1)]}\"\n"
f"{'':8}return {metrics}\n"
f"{'':4}else:\n"
- f"{'':8}target_values = {target_values}\n"
- f"{'':8}output_table = pd.DataFrame({{'{metrics}': np.array(target_values)[np.argmax(prediction.iloc[0:, 1:].values, axis=1)]}})\n"
+ f"{'':8}output_table = pd.DataFrame({{'{metrics}': np.where(prediction[prediction.columns[{target_index+1}]] > {threshold}, '{target_values[target_index]}', '{target_values[abs(target_index-1)]}')}})\n"
f"{'':8}return output_table"
)
"""
@@ -1317,21 +1339,20 @@ def _binary_target(
Classification = "B"
return Classification
else:
- target_values = ['A', 'B']
- output_table = pd.DataFrame({'Classification': np.array(target_values)[np.argmax(prediction.iloc[0:, 1:].values, axis=1)]})
+ output_table = pd.DataFrame({'Classification': np.where(prediction[prediction.columns[2]] > .5, 'B', 'A')})
return output_table
"""
# One return that is the classification
elif len(returns) == 1 and returns[0]:
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction\n"
+ f"{'':8}return prediction[0]\n"
f"{'':4}else:\n"
f"{'':8}return pd.DataFrame({{'{metrics}': prediction}})"
)
"""
if input_array.shape[0] == 1:
- return prediction
+ return prediction[0]
else:
return pd.DataFrame({'Classification': prediction})
"""
@@ -1339,45 +1360,46 @@ def _binary_target(
elif len(returns) == 1 and not returns[0]:
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}if prediction > {threshold}:\n"
- f"{'':12}{metrics} = \"{target_values[0]}\"\n"
+ f"{'':8}if prediction[0] > {threshold}:\n"
+ f"{'':12}{metrics} = \"{target_values[target_index]}\"\n"
f"{'':8}else:\n"
- f"{'':12}{metrics} = \"{target_values[1]}\"\n"
+ f"{'':12}{metrics} = \"{target_values[abs(target_index-1)]}\"\n"
f"{'':8}return {metrics}\n"
f"{'':4}else:\n"
- f"{'':8}return pd.DataFrame({{'{metrics}': ['{target_values[0]}' if p > {threshold} else '{target_values[1]}' for p in prediction]}})\n"
+ f"{'':8}return pd.DataFrame({{'{metrics}': ['{target_values[target_index]}' if p > {threshold} else '{target_values[abs(target_index-1)]}' for p in prediction]}})\n"
)
"""
if input_array.shape[0] == 1:
- if prediction > 0.5:
+ if prediction[0] > 0.5:
Classification = "A"
else:
Classification = "B"
return Classification
else:
- return pd.DataFrame({'Classification': ['A' if p > 0.5 else 'B' for p in prediction]})
+ return pd.DataFrame({'Classification': ['B' if p > 0.5 else 'A' for p in prediction]})
"""
# Two returns from the prediction method
elif len(returns) == 2 and sum(returns) == 0:
# Only probabilities returned; return classification for larger value
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}if prediction[0] > prediction[1]:\n"
- f"{'':12}{metrics} = \"{target_values[0]}\"\n"
+ f"{'':8}if prediction[0][{target_index}] > {threshold}:\n"
+ f"{'':12}{metrics} = \"{target_values[target_index]}\"\n"
f"{'':8}else:\n"
- f"{'':12}{metrics} = \"{target_values[1]}\"\n\n"
+ f"{'':12}{metrics} = \"{target_values[abs(target_index-1)]}\"\n\n"
f"{'':8}return {metrics}\n"
f"{'':4}else:\n"
f"{'':8}target_values = {target_values}\n"
- f"{'':8}output_table = pd.DataFrame({{'{metrics}' : np.array(target_values)[np.argmax(prediction, axis=1)]}})\n"
+ f"{'':8}prediction = pd.DataFrame(prediction)\n"
+ f"{'':8}output_table = pd.DataFrame({{'{metrics}': np.where(prediction[prediction.columns[{target_index}]] > {threshold}, '{target_values[target_index]}', '{target_values[abs(target_index-1)]}')}})\n"
f"{'':8}return output_table"
)
"""
if input_array.shape[0] == 1:
- if prediction[0] > prediction[1]:
- Classification = "A"
- else:
+ if prediction[0][0] > .5:
Classification = "B"
+ else:
+ Classification = "A"
return Classification
else:
@@ -1387,18 +1409,19 @@ def _binary_target(
"""
# Classification and probability returned; return classification value
elif len(returns) > 1 and sum(returns) == 1:
+ # TODO: Either figure out how to handle threshold or add warning
# Determine which return is the classification value
class_index = [i for i, x in enumerate(returns) if x][0]
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}{metrics} = prediction[{class_index}]\n{'':8}return {metrics}\n"
+ f"{'':8}{metrics} = prediction[0][{class_index}]\n{'':8}return {metrics}\n"
f"{'':4}else:\n"
f"{'':8}output_table = pd.DataFrame({{'{metrics}': [p[{class_index}] for p in prediction]}})\n"
f"{'':8}return output_table"
)
"""
if input_array.shape[0] == 1:
- Classification = prediction[1]
+ Classification = prediction[0][1]
return Classification
else:
output_table = pd.DataFrame({'Classification': [p[1] for p in prediction]})
@@ -1407,7 +1430,6 @@ def _binary_target(
else:
cls._invalid_predict_config()
elif len(metrics) == 2:
- # TODO: change to align with other cases and assign target_values to classification column
# H2O models with two metrics are assumed to be classification + probability
if h2o_model:
warn(
@@ -1417,19 +1439,30 @@ def _binary_target(
)
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction[1][0], float(prediction[1][2])\n"
+ f"{'':8}if prediction[1][{target_index+1}] > {threshold}:\n"
+ f"{'':12}{metrics[0]} = '{target_values[target_index]}'\n"
+ f"{'':8}else:\n"
+ f"{'':12}{metrics[0]} = '{target_values[abs(target_index-1)]}'\n"
+ f"{'':8}return {metrics[0]}, float(prediction[1][{target_index+1}])\n"
f"{'':4}else:\n"
- f"{'':8}output_table = prediction.drop(prediction.columns[1], axis=1)\n"
+ f"{'':8}output_table = prediction.drop(prediction.columns[{abs(target_index-1)+1}], axis=1)\n"
+ f"{'':8}classifications = np.where(prediction[prediction.columns[{target_index+1}]] > {threshold}, '{target_values[target_index]}', '{target_values[abs(target_index-1)]}')\n"
f"{'':8}output_table.columns = {metrics}\n"
+ f"{'':8}output_table['{metrics[0]}'] = classifications\n"
f"{'':8}return output_table"
)
"""
if input_array.shape[0] == 1:
- return prediction[1][0], float(prediction[1][2])
+ if prediction[1][1] > 0.5:
+ Classification = '1'
+ else:
+ Classification = '0'
+ return EM_CLASSIFICATION, float(prediction[1][1])
else:
- output_table = prediction.drop(prediction.columns[1], axis=1)
- output_table.columns = ['Classification', 'Probability']
- return output_table
+ output_table = prediction.drop(prediction.columns[2], axis=1)
+ classifications = np.where(prediction[prediction.columns[1]] > 0.5, '0', '1')
+ output_table.columns = ['EM_CLASSIFICATION', 'EM_EVENTPROBABILITY']
+ output_table['EM_CLASSIFICATION'] = classifications
"""
# Calculate the classification; return the classification and probability
elif sum(returns) == 0 and len(returns) == 1:
@@ -1440,25 +1473,25 @@ def _binary_target(
)
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}if prediction > {threshold}:\n"
- f"{'':12}{metrics[0]} = \"{target_values[0]}\"\n"
+ f"{'':8}if prediction[0] > {threshold}:\n"
+ f"{'':12}{metrics[0]} = \"{target_values[target_index]}\"\n"
f"{'':8}else:\n"
- f"{'':12}{metrics[0]} = \"{target_values[1]}\"\n\n"
- f"{'':8}return {metrics[0]}, prediction\n"
+ f"{'':12}{metrics[0]} = \"{target_values[abs(target_index-1)]}\"\n\n"
+ f"{'':8}return {metrics[0]}, prediction[0]\n"
f"{'':4}else:\n"
- f"{'':8}classifications = ['{target_values[0]}' if p > {threshold} else '{target_values[1]}' for p in prediction]\n"
+ f"{'':8}classifications = ['{target_values[target_index]}' if p > {threshold} else '{target_values[abs(target_index-1)]}' for p in prediction]\n"
f"{'':8}return pd.DataFrame({{'{metrics[0]}': classifications, '{metrics[1]}': prediction}})"
)
"""
if input_array.shape[0] == 1:
- if prediction > 0.5:
- Classification = "A"
- else:
+ if prediction[0] > 0.5:
Classification = "B"
+ else:
+ Classification = "A"
- return Classification, prediction
+ return Classification, prediction[0]
else:
- classifications = ['A' if p > 0.5 else 'B' for p in prediction]
+ classifications = ['B' if p > 0.5 else 'A' for p in prediction]
return pd.DataFrame({'Classification': classifications, 'Probability': prediction})
"""
# Calculate the classification; return the classification and probability
@@ -1470,41 +1503,42 @@ def _binary_target(
)
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}if prediction[0] > prediction[1]:\n"
- f"{'':12}{metrics[0]} = \"{target_values[0]}\"\n"
+ f"{'':8}if prediction[0][{target_index}] > {threshold}:\n"
+ f"{'':12}{metrics[0]} = \"{target_values[target_index]}\"\n"
f"{'':8}else:\n"
- f"{'':12}{metrics[0]} = \"{target_values[1]}\"\n"
- f"{'':8}return {metrics[0]}, prediction[0]\n"
+ f"{'':12}{metrics[0]} = \"{target_values[abs(target_index-1)]}\"\n"
+ f"{'':8}return {metrics[0]}, prediction[0][{target_index}]\n"
f"{'':4}else:\n"
f"{'':8}df = pd.DataFrame(prediction)\n"
- f"{'':8}proba = df[0]\n"
- f"{'':8}classifications = np.where(df[0] > df[1], '{target_values[0]}', '{target_values[1]}')\n"
+ f"{'':8}proba = df[{target_index}]\n"
+ f"{'':8}classifications = np.where(df[{target_index}] > {threshold}, '{target_values[target_index]}', '{target_values[abs(target_index-1)]}')\n"
f"{'':8}return pd.DataFrame({{'{metrics[0]}': classifications, '{metrics[1]}': proba}})"
)
"""
if input_array.shape[0] == 1:
- if prediction[0] > prediction[1]:
- Classification = "A"
- else:
+ if prediction[0][1] > .5:
Classification = "B"
- return Classification, prediction[0]
+ else:
+ Classification = "A"
+ return Classification, prediction[0][1]
else:
df = pd.DataFrame(prediction)
proba = df[0]
- classifications = np.where(df[0] > df[1], 'A', 'B')
+ classifications = np.where(df[1] > .5, 'B', 'A')
return pd.DataFrame({'Classification': classifications, 'Probability': proba})
"""
+ # TODO: Potentially add threshold
# Return classification and probability value
elif sum(returns) == 1 and len(returns) == 2:
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction[0], prediction[1]\n"
+ f"{'':8}return prediction[0][0], prediction[0][1]\n"
f"{'':4}else:\n"
f"{'':8}return pd.DataFrame(prediction, columns={metrics})"
)
"""
if input_array.shape[0] == 1:
- return prediction[0], prediction[1]
+ return prediction[0][0], prediction[0][1]
else:
return pd.DataFrame(prediction, columns=['Classification', 'Probability'])
"""
@@ -1517,38 +1551,42 @@ def _binary_target(
# Determine which return is the classification value
class_index = [i for i, x in enumerate(returns) if x][0]
if class_index == 0:
- metric_list = '"' + '","'.join(metrics) + '","drop"'
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction[0], prediction[1]\n"
+ f"{'':8}return prediction[0][0], prediction[0][{target_index+1}]\n"
f"{'':4}else:\n"
- f"{'':8}output_table = pd.DataFrame(prediction, columns=[{metric_list}])\n"
- f"{'':8}return output_table.drop('drop', axis=1)"
+ f"{'':8}prediction = pd.DataFrame(prediction)\n"
+ f"{'':8}output_table = prediction.drop(prediction.columns[{abs(target_index-1)+1}], axis=1)\n"
+ f"{'':8}output_table.columns = {metrics}\n"
+ f"{'':8}return output_table"
)
"""
if input_array.shape[0] == 1:
- return prediction[0], prediction[1]
+ return prediction[0][0], prediction[0][2]
else:
- output_table = pd.DataFrame(prediction, columns=["Classification","Probability","drop"])
- return output_table.drop('drop', axis=1)
+ output_table = prediction.drop(prediction.columns[1], axis=1)
+ output_table.columns = ["Classification", "Probability"]
+ return output_table
"""
else:
- metric_list = '"' + '","drop","'.join(metrics[::-1]) + '"'
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction[{class_index}], prediction[0]\n"
+ f"{'':8}return prediction[0][{class_index}], prediction[0][{target_index}]\n"
f"{'':4}else:\n"
- f"{'':8}output_table = pd.DataFrame(prediction, columns=[{metric_list}])\n"
+ f"{'':8}prediction = pd.DataFrame(prediction)\n"
+ f"{'':8}output_table = prediction.drop(prediction.columns[{abs(target_index-1)}], axis=1)\n"
f"{'':8}output_table = output_table[output_table.columns[::-1]]\n"
- f"{'':8}return output_table.drop('drop', axis=1)"
+ f"{'':8}output_table.columns = {metrics}\n"
+ f"{'':8}return output_table"
)
"""
if input_array.shape[0] == 1:
- return prediction[2], prediction[0]
+ return prediction[0][2], prediction[0][0]
else:
- output_table = pd.DataFrame(prediction, columns=["Probability","drop","Classification"])
+ output_table = prediction.drop(prediction.columns[0], axis=1)
output_table = output_table[output_table.columns[::-1]]
+ output_table.columns = ["Classification", "Probability"]
return output_table.drop('drop', axis=1)
"""
else:
@@ -1578,59 +1616,63 @@ def _binary_target(
)
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}if prediction > {threshold}:\n"
- f"{'':12}{metrics[0]} = \"{target_values[0]}\"\n"
+ f"{'':8}if prediction[0] > {threshold}:\n"
+ f"{'':12}{metrics[0]} = \"{target_values[target_index]}\"\n"
f"{'':8}else:\n"
- f"{'':12}{metrics[0]} = \"{target_values[1]}\"\n"
- f"{'':8}return {metrics[0]}, prediction, 1 - prediction\n"
+ f"{'':12}{metrics[0]} = \"{target_values[abs(target_index-1)]}\"\n"
+ f"{'':8}return {metrics[0]}, prediction[0], 1 - prediction[0]\n"
f"{'':4}else:\n"
- f"{'':8}classifications = ['{target_values[0]}' if p > {threshold} else '{target_values[1]}' for p in prediction]\n"
+ f"{'':8}classifications = ['{target_values[target_index]}' if p > {threshold} else '{target_values[abs(target_index-1)]}' for p in prediction]\n"
f"{'':8}output_table = pd.DataFrame({{'{metrics[0]}': classifications, '{metrics[1]}': prediction}})\n"
f"{'':8}output_table['{metrics[2]}'] = 1 - output_table['{metrics[1]}']\n"
f"{'':8}return output_table"
)
"""
if input_array.shape[0] == 1:
- if prediction > 0.5:
- Classification = "A"
- else:
+ if prediction[0] > 0.5:
Classification = "B"
- return Classification, prediction, 1 - prediction
+ else:
+ Classification = "A"
+ return Classification, prediction[0], 1 - prediction[0]
else:
- classifications = ['A' if p > 0.5 else 'B' for p in prediction]
+ classifications = ['B' if p > 0.5 else 'A' for p in prediction]
output_table = pd.DataFrame({'Classification': classifications, 'Proba_0': prediction})
output_table['Proba_1'] = 1 - output_table['Proba_0']
return output_table
"""
elif sum(returns) == 0 and len(returns) == 2:
+ # TODO: Make decision on whether ordering should follow given pattern or reflect input ordering
warn(
"Due to the ambiguity of the provided metrics and prediction return"
" types, the score code assumes the return order to be: "
- "[classification, probability of event, probability of no event]."
+ "[classification, probability of event, probability of no event] "
+ "for a single return. For batch scoring, the return order of the "
+ "probabilities will mirror their return order in the model."
)
- metric_list = '"' + '","'.join(metrics[1:]) + '"'
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}if prediction[0] > prediction[1]:\n"
- f"{'':12}{metrics[0]} = \"{target_values[0]}\"\n"
+ f"{'':8}if prediction[0][{target_index}] > {threshold}:\n"
+ f"{'':12}{metrics[0]} = \"{target_values[target_index]}\"\n"
f"{'':8}else:\n"
- f"{'':12}{metrics[0]} = \"{target_values[1]}\"\n"
- f"{'':8}return {metrics[0]}, prediction[0], prediction[1]\n"
+ f"{'':12}{metrics[0]} = \"{target_values[abs(target_index-1)]}\"\n"
+ f"{'':8}return {metrics[0]}, prediction[0][{target_index}], prediction[0][{abs(target_index-1)}]\n"
f"{'':4}else:\n"
- f"{'':8}output_table = pd.DataFrame(prediction, columns=[{metric_list}])\n"
- f"{'':8}output_table.insert(0, '{metrics[0]}', np.array({target_values})[np.argmax(output_table.values, axis=1)])\n"
+ f"{'':8}output_table = pd.DataFrame(prediction, columns={metrics[1:]})\n"
+ f"{'':8}classifications = np.where(output_table[output_table.columns[{target_index}]] > {threshold}, '{target_values[target_index]}', '{target_values[abs(target_index-1)]}')\n"
+ f"{'':8}output_table.insert(loc=0, column='{metrics[0]}', value=classifications)\n"
f"{'':8}return output_table"
)
"""
if input_array.shape[0] == 1:
- if prediction[0] > prediction[1]:
+ if prediction[0][0] > prediction[0][1]:
Classification = "A"
else:
Classification = "B"
- return Classification, prediction[0], prediction[1]
+ return Classification, prediction[0][0], prediction[0][1]
else:
output_table = pd.DataFrame(prediction, columns=["Proba_0","Proba_1"])
- output_table.insert(0, 'Classification', np.array(['A', 'B'])[np.argmax(output_table.values, axis=1)])
+ classifications = np.where(prediction[prediction.columns[2]] > .5, 'B', 'A')
+ output_table.insert(loc=0, column='Classification', value=classifications)
return output_table
"""
# Find which return is the classification, then return probabilities
@@ -1638,37 +1680,35 @@ def _binary_target(
# Determine which return is the classification value
class_index = [i for i, x in enumerate(returns) if x][0]
if class_index == 0:
- metric_list = '"' + '","'.join(metrics[:2]) + '"'
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction[0], prediction[1], 1 - prediction[1]\n"
+ f"{'':8}return prediction[0][0], prediction[0][1], 1 - prediction[0][1]\n"
f"{'':4}else:\n"
- f"{'':8}output_table = pd.DataFrame(prediction, columns=[{metric_list}])\n"
+ f"{'':8}output_table = pd.DataFrame(prediction, columns={metrics[:2]})\n"
f"{'':8}output_table['{metrics[2]}'] = 1 - output_table['{metrics[1]}']\n"
f"{'':8}return output_table"
)
"""
if input_array.shape[0] == 1:
- return prediction[0], prediction[1], 1 - prediction[1]
+ return prediction[0][0], prediction[0][1], 1 - prediction[0][1]
else:
output_table = pd.DataFrame(prediction, columns=["Classification","Proba_0"])
output_table['Proba_1'] = 1 - output_table['Proba_0']
return output_table
"""
else:
- metric_list = '"' + '","'.join(metrics[1::-1]) + '"'
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction[1], prediction[0], 1 - prediction[0]\n"
+ f"{'':8}return prediction[0][1], prediction[0][0], 1 - prediction[0][0]\n"
f"{'':4}else:\n"
- f"{'':8}output_table = pd.DataFrame(prediction, columns=[{metric_list}])\n"
+ f"{'':8}output_table = pd.DataFrame(prediction, columns={metrics[1::-1]})\n"
f"{'':8}output_table = output_table[output_table.columns[::-1]]\n"
f"{'':8}output_table['{metrics[2]}'] = 1 - output_table['{metrics[1]}']\n"
f"{'':8}return output_table"
)
"""
if input_array.shape[0] == 1:
- return prediction[1], prediction[0], 1 - prediction[0]
+ return prediction[0][1], prediction[0][0], 1 - prediction[0][0]
else:
output_table = pd.DataFrame(prediction, columns=["Proba_0","Classification"])
output_table = output_table[output_table.columns[::-1]]
@@ -1679,13 +1719,13 @@ def _binary_target(
elif sum(returns) == 1 and len(returns) == 3:
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction[0], prediction[1], prediction[2]\n"
+ f"{'':8}return prediction[0][0], prediction[0][1], prediction[0][2]\n"
f"{'':4}else:\n"
f"{'':8}return pd.DataFrame(prediction, columns={metrics})"
)
"""
if input_array.shape[0] == 1:
- return prediction[0], prediction[1], prediction[2]
+ return prediction[0][0], prediction[0][1], prediction[0][2]
else:
return pd.DataFrame(prediction, columns=['Classification', 'Proba_0', 'Proba_1'])
"""
@@ -1752,13 +1792,13 @@ def _nonbinary_targets(
elif len(returns) == 1:
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction\n"
+ f"{'':8}return prediction[0][0]\n"
f"{'':4}else:\n"
f"{'':8}return pd.DataFrame({{'{metrics}': prediction}})"
)
"""
if input_array.shape[0] == 1:
- return prediction
+ return prediction[0]
else:
return pd.DataFrame({'Classification': prediction})
"""
@@ -1766,7 +1806,7 @@ def _nonbinary_targets(
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
f"{'':8}target_values = {target_values}\n"
- f"{'':8}return target_values[prediction.index(max(prediction))]\n"
+ f"{'':8}return target_values[prediction[0].index(max(prediction[0]))]\n"
f"{'':4}else:\n"
f"{'':8}output_table = pd.DataFrame({{'{metrics}' : np.array({target_values})[np.argmax(prediction, axis=1)]}})\n"
f"{'':8}return output_table"
@@ -1774,7 +1814,7 @@ def _nonbinary_targets(
"""
if input_array.shape[0] == 1:
target_values = ['A', 'B', 'C']
- return target_values[prediction.index(max(prediction))]
+ return target_values[prediction[0].index(max(prediction[0]))]
else:
output_table = pd.DataFrame({'Classification' : np.array(['A', 'B', 'C'])[np.argmax(prediction, axis=1)]})
return output_table
@@ -1784,13 +1824,13 @@ def _nonbinary_targets(
class_index = [i for i, x in enumerate(returns) if x][0]
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction[{class_index}]\n"
+ f"{'':8}return prediction[0][{class_index}]\n"
f"{'':4}else:\n"
f"{'':8}return pd.DataFrame({{'{metrics}': [p[{class_index}] for p in prediction]}})"
)
"""
if input_array.shape[0] == 1:
- return prediction[0]
+ return prediction[0][0]
else:
return pd.DataFrame({'Classification': [p[0] for p in prediction]})
"""
@@ -1821,8 +1861,8 @@ def _nonbinary_targets(
cls.score_code += (
f"{'':4}target_values = {target_values}\n"
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return target_values[prediction.index(max(prediction))], "
- f"max(prediction)\n"
+ f"{'':8}return target_values[prediction[0].index(max(prediction[0]))], "
+ f"max(prediction[0])\n"
f"{'':4}else:\n"
f"{'':8}df = pd.DataFrame(prediction)\n"
f"{'':8}index = np.argmax(df.values, axis=1)\n"
@@ -1833,7 +1873,7 @@ def _nonbinary_targets(
"""
target_values = ['A', 'B', 'C']
if input_array.shape[0] == 1:
- return target_values[prediction.index(max(prediction))], max(prediction)
+ return target_values[prediction[0].index(max(prediction[0]))], max(prediction[0])
else:
df = pd.DataFrame(prediction)
index = np.argmax(df.values, axis=1)
@@ -1846,8 +1886,8 @@ def _nonbinary_targets(
class_index = [i for i, x in enumerate(returns) if x][0]
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return prediction[{class_index}], "
- f"max(prediction[:{class_index}] + prediction[{class_index + 1}:])\n"
+ f"{'':8}return prediction[0][{class_index}], "
+ f"max(prediction[0][:{class_index}] + prediction[0][{class_index + 1}:])\n"
f"{'':4}else:\n"
f"{'':8}df = pd.DataFrame(prediction)\n"
f"{'':8}probas = df.drop({class_index}, axis=1)\n"
@@ -1856,7 +1896,7 @@ def _nonbinary_targets(
)
"""
if input_array.shape[0] == 1:
- return prediction[0], max(prediction[:0] + prediction[1:])
+ return prediction[0][0], max(prediction[0][:0] + prediction[0][1:])
else:
df = pd.DataFrame(prediction)
probas = df.drop(0, axis=1)
@@ -1907,7 +1947,7 @@ def _nonbinary_targets(
len(metrics) == (len(target_values) + 1) == len(returns)
and sum(returns) == 1
):
- proba_returns = [f"prediction[{i}]" for i in range(len(returns))]
+ proba_returns = [f"prediction[0][{i}]" for i in range(len(returns))]
cls.score_code += (
f"{'':4}if input_array.shape[0] == 1:\n"
f"{'':8}return {', '.join(proba_returns)}\n"
@@ -1917,7 +1957,7 @@ def _nonbinary_targets(
)
"""
if input_array.shape[0] == 1:
- return prediction[0], prediction[1], prediction[2]
+ return prediction[0][0], prediction[0][1], prediction[0][2]
else:
output_table = pd.DataFrame(prediction, columns=['Proba_0', 'Proba_1', 'Proba_2'])
return output_table
@@ -1925,11 +1965,11 @@ def _nonbinary_targets(
elif (len(metrics) - 1) == len(returns) == len(target_values) and sum(
returns
) == 0:
- proba_returns = [f"prediction[{i}]" for i in range(len(returns))]
+ proba_returns = [f"prediction[0][{i}]" for i in range(len(returns))]
cls.score_code += (
f"{'':4}target_values = {target_values}\n\n"
f"{'':4}if input_array.shape[0] == 1:\n"
- f"{'':8}return target_values[prediction.index(max(prediction))], "
+ f"{'':8}return target_values[prediction[0].index(max(prediction[0]))], "
f"{', '.join(proba_returns)}\n"
f"{'':4}else:\n"
f"{'':8}output_table = pd.DataFrame(prediction, columns={metrics[1:]})\n"
@@ -1941,7 +1981,7 @@ def _nonbinary_targets(
target_values = ['A', 'B', 'C']
if input_array.shape[0] == 1:
- return target_values[prediction.index(max(prediction))], prediction[0], prediction[1], prediction[2]
+ return target_values[prediction[0].index(max(prediction[0]))], prediction[0][0], prediction[0][1], prediction[0][2]
else:
output_table = pd.DataFrame(prediction, columns=['Proba_0', 'Proba_1', 'Proba_2'])
classifications = np.array(target_values)[np.argmax(output_table.values, axis=1)]
diff --git a/tests/unit/test_write_score_code.py b/tests/unit/test_write_score_code.py
index f94a122c..c9f35715 100644
--- a/tests/unit/test_write_score_code.py
+++ b/tests/unit/test_write_score_code.py
@@ -331,7 +331,7 @@ def test_single_metric(self):
self.sc._no_targets_no_thresholds(metrics, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [0.5]
+ prediction = [[0.5]]
self.assertEqual(self.execute_snippet(input_array, prediction), 0.5)
# Multi row
input_array = pd.DataFrame({"A": [0.9, 1, 1.1]})
@@ -369,7 +369,7 @@ def test_multi_metric(self):
self.sc._no_targets_no_thresholds(metrics, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = ["i", 0.3, 0.4, 0.5]
+ prediction = [["i", 0.3, 0.4, 0.5]]
self.assertEqual(
self.execute_snippet(input_array, prediction), ("i", 0.3, 0.4, 0.5)
)
@@ -449,7 +449,7 @@ def test_one_metric_h2o(self):
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
prediction = [[], [1, 2, 3]]
- self.assertEqual(self.execute_snippet(input_array, prediction), "A")
+ self.assertEqual(self.execute_snippet(input_array, prediction), "B")
# Multi row
input_array = pd.DataFrame({"A": [0, 1]})
prediction = pd.DataFrame(
@@ -469,7 +469,7 @@ def test_one_metric_one_return_classification(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = 0.5
+ prediction = [0.5]
self.assertEqual(self.execute_snippet(input_array, prediction), 0.5)
# Multi row
input_array = pd.DataFrame({"A": [0.9, 1, 1.1]})
@@ -488,14 +488,14 @@ def test_one_metric_one_return_probability(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = 1
- self.assertEqual(self.execute_snippet(input_array, prediction), "A")
+ prediction = [1]
+ self.assertEqual(self.execute_snippet(input_array, prediction), "B")
# Multi row
input_array = pd.DataFrame({"A": [1, 0, 1]})
prediction = [0, 1, 0]
pd.testing.assert_frame_equal(
self.execute_snippet(input_array, prediction),
- pd.DataFrame({metrics: ["B", "A", "B"]}),
+ pd.DataFrame({metrics: ["A", "B", "A"]}),
)
def test_one_metric_two_returns(self):
@@ -509,7 +509,7 @@ def test_one_metric_two_returns(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [1, 0]
+ prediction = [[1, 0]]
self.assertEqual(self.execute_snippet(input_array, prediction), "A")
# Multi row
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -528,7 +528,7 @@ def test_one_metric_three_returns(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [0, "Y", "z"]
+ prediction = [[0, "Y", "z"]]
self.assertEqual(self.execute_snippet(input_array, prediction), "Y")
# Multi row
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -542,13 +542,15 @@ def test_two_metrics_h2o(self):
metrics = ["Classification", "Probability"]
returns = ["", int, int]
self.sc.score_code += (
- "import pandas as pd\n" "def test_snippet(input_array, prediction):\n"
+ "import pandas as pd\n"
+ "import numpy as np\n"
+ "def test_snippet(input_array, prediction):\n"
)
self.sc._binary_target(metrics, self.target_values, returns, h2o_model=True)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
prediction = [[], ["a", -1, 1]]
- self.assertEqual(self.execute_snippet(input_array, prediction), ("a", 1))
+ self.assertEqual(self.execute_snippet(input_array, prediction), ("B", 1.0))
# Multi row
input_array = pd.DataFrame({"A": [0, 1]})
prediction = pd.DataFrame(
@@ -556,7 +558,7 @@ def test_two_metrics_h2o(self):
)
pd.testing.assert_frame_equal(
self.execute_snippet(input_array, prediction),
- pd.DataFrame({"Classification": [0, 1], "Probability": [0.1, 0.8]}),
+ pd.DataFrame({"Classification": ["A", "B"], "Probability": [0.1, 0.8]}),
)
def test_two_metrics_one_return(self):
@@ -568,14 +570,14 @@ def test_two_metrics_one_return(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = 0.2
- self.assertEqual(self.execute_snippet(input_array, prediction), ("B", 0.2))
+ prediction = [0.2]
+ self.assertEqual(self.execute_snippet(input_array, prediction), ("A", 0.2))
# Multi row
input_array = pd.DataFrame({"A": [1, 0, 1]})
prediction = [1, -1]
pd.testing.assert_frame_equal(
self.execute_snippet(input_array, prediction),
- pd.DataFrame({"Classification": ["A", "B"], "Probability": [1, -1]}),
+ pd.DataFrame({"Classification": ["B", "A"], "Probability": [1, -1]}),
)
def test_two_metrics_two_returns_no_classification(self):
@@ -589,14 +591,14 @@ def test_two_metrics_two_returns_no_classification(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [0.2, 0.8]
- self.assertEqual(self.execute_snippet(input_array, prediction), ("B", 0.2))
+ prediction = [[0.2, 0.8]]
+ self.assertEqual(self.execute_snippet(input_array, prediction), ("B", 0.8))
# Multi row
input_array = pd.DataFrame({"A": [1, 0, 1]})
prediction = [[0.9, 0.1], [0.4, 0.6]]
pd.testing.assert_frame_equal(
self.execute_snippet(input_array, prediction),
- pd.DataFrame({"Classification": ["A", "B"], "Probability": [0.9, 0.4]}),
+ pd.DataFrame({"Classification": ["A", "B"], "Probability": [0.1, 0.6]}),
)
def test_two_metrics_two_returns_classification(self):
@@ -610,7 +612,7 @@ def test_two_metrics_two_returns_classification(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = ["B", 0.2]
+ prediction = [["B", 0.2]]
self.assertEqual(self.execute_snippet(input_array, prediction), ("B", 0.2))
# Multi row
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -631,14 +633,14 @@ def test_two_metrics_three_returns_class_first(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = ["B", 0.2, 0.8]
- self.assertEqual(self.execute_snippet(input_array, prediction), ("B", 0.2))
+ prediction = [["B", 0.2, 0.8]]
+ self.assertEqual(self.execute_snippet(input_array, prediction), ("B", 0.8))
# Multi row
input_array = pd.DataFrame({"A": [1, 0, 1]})
prediction = [["A", 0.9, 0.1], ["B", 0.4, 0.6]]
pd.testing.assert_frame_equal(
self.execute_snippet(input_array, prediction),
- pd.DataFrame({"Classification": ["A", "B"], "Probability": [0.9, 0.4]}),
+ pd.DataFrame({"Classification": ["A", "B"], "Probability": [0.1, 0.6]}),
)
def test_two_metrics_three_returns_class_last(self):
@@ -652,14 +654,14 @@ def test_two_metrics_three_returns_class_last(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [0.2, 0.8, "B"]
- self.assertEqual(self.execute_snippet(input_array, prediction), ("B", 0.2))
+ prediction = [[0.2, 0.8, "B"]]
+ self.assertEqual(self.execute_snippet(input_array, prediction), ("B", 0.8))
# Multi row
input_array = pd.DataFrame({"A": [1, 0, 1]})
prediction = [[0.9, 0.1, "A"], [0.4, 0.6, "B"]]
pd.testing.assert_frame_equal(
self.execute_snippet(input_array, prediction),
- pd.DataFrame({"Classification": ["A", "B"], "Probability": [0.9, 0.4]}),
+ pd.DataFrame({"Classification": ["A", "B"], "Probability": [0.1, 0.6]}),
)
def test_three_metrics_h2o(self):
@@ -695,9 +697,9 @@ def test_three_metrics_one_return(self):
print(self.sc.score_code)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = 0.9
+ prediction = [0.9]
self.assertEqual(
- self.execute_snippet(input_array, prediction), ("A", 0.9, 1 - 0.9)
+ self.execute_snippet(input_array, prediction), ("B", 0.9, 1 - 0.9)
)
# Multiple rows
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -706,7 +708,7 @@ def test_three_metrics_one_return(self):
self.execute_snippet(input_array, prediction),
pd.DataFrame(
{
- "Classification": ["A", "B"],
+ "Classification": ["B", "A"],
"Proba_0": [0.9, 0.1],
"Proba_1": [1 - 0.9, 1 - 0.1],
}
@@ -724,8 +726,8 @@ def test_three_metrics_two_returns_no_class(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [0.9, 0.1]
- self.assertEqual(self.execute_snippet(input_array, prediction), ("A", 0.9, 0.1))
+ prediction = [[0.9, 0.1]]
+ self.assertEqual(self.execute_snippet(input_array, prediction), ("A", 0.1, 0.9))
# Multiple rows
input_array = pd.DataFrame({"A": [1, 0, 1]})
prediction = [[0.9, 0.1], [0.2, 0.8]]
@@ -751,7 +753,7 @@ def test_three_metrics_two_returns_class_first(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = ["A", 0.9]
+ prediction = [["A", 0.9]]
self.assertEqual(
self.execute_snippet(input_array, prediction), ("A", 0.9, 1 - 0.9)
)
@@ -780,7 +782,7 @@ def test_three_metrics_two_returns_class_last(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [0.9, "A"]
+ prediction = [[0.9, "A"]]
self.assertEqual(
self.execute_snippet(input_array, prediction), ("A", 0.9, 1 - 0.9)
)
@@ -809,7 +811,7 @@ def test_three_metrics_three_returns(self):
self.sc._binary_target(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = ["A", 0.9, 0.1]
+ prediction = [["A", 0.9, 0.1]]
self.assertEqual(self.execute_snippet(input_array, prediction), ("A", 0.9, 0.1))
# Multiple rows
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -869,7 +871,7 @@ def test_one_metric_one_return(self):
self.sc._nonbinary_targets(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = "C"
+ prediction = ["C"]
self.assertEqual(self.execute_snippet(input_array, prediction), "C")
# Multiple rows
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -885,7 +887,7 @@ def test_one_metric_probability_returns(self):
self.sc._nonbinary_targets(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [0.1, 0.2, 0.3]
+ prediction = [[0.1, 0.2, 0.3]]
self.assertEqual(self.execute_snippet(input_array, prediction), "C")
# Multiple rows
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -901,7 +903,7 @@ def test_one_metric_classification_and_probability_returns(self):
self.sc._nonbinary_targets(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = ["C", 0.1, 0.2, 0.3]
+ prediction = [["C", 0.1, 0.2, 0.3]]
self.assertEqual(self.execute_snippet(input_array, prediction), "C")
# Multiple rows
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -935,7 +937,7 @@ def test_two_metrics_return_probabilities(self):
self.sc._nonbinary_targets(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [0.1, 0.2, 0.3]
+ prediction = [[0.1, 0.2, 0.3]]
self.assertEqual(self.execute_snippet(input_array, prediction), ("C", 0.3))
# Multiple rows
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -951,7 +953,7 @@ def test_two_metrics_return_classification_and_probability(self):
self.sc._nonbinary_targets(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = ["C", 0.1, 0.2, 0.3]
+ prediction = [["C", 0.1, 0.2, 0.3]]
self.assertEqual(self.execute_snippet(input_array, prediction), ("C", 0.3))
# Multiple rows
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -1014,7 +1016,7 @@ def test_return_all_probabilities(self):
self.sc._nonbinary_targets(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [0.1, 0.2, 0.3]
+ prediction = [[0.1, 0.2, 0.3]]
self.assertEqual(self.execute_snippet(input_array, prediction), (0.1, 0.2, 0.3))
# Multiple rows
input_array = pd.DataFrame({"A": [1, 0, 1]})
@@ -1032,7 +1034,7 @@ def test_return_all_probabilities_and_classification(self):
self.sc._nonbinary_targets(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = ["P", 0.1, 0.2, 0.3]
+ prediction = [["P", 0.1, 0.2, 0.3]]
self.assertEqual(
self.execute_snippet(input_array, prediction), ("P", 0.1, 0.2, 0.3)
)
@@ -1057,7 +1059,7 @@ def test_return_all_probabilities_generate_classification(self):
self.sc._nonbinary_targets(metrics, self.target_values, returns)
# Single row
input_array = pd.DataFrame([[1]], columns=["A"], index=[0])
- prediction = [0.1, 0.2, 0.3]
+ prediction = [[0.1, 0.2, 0.3]]
self.assertEqual(
self.execute_snippet(input_array, prediction), ("C", 0.1, 0.2, 0.3)
)
@@ -1103,7 +1105,7 @@ def test_predictions_to_metrics():
metrics = ["Classification", "Probability"]
target_values = ["1", "0"]
sc._predictions_to_metrics(metrics, returns, target_values)
- func.assert_called_once_with(metrics, ["1", "0"], returns, None, False)
+ func.assert_called_once_with(metrics, ["1", "0"], returns, None, 1, False)
with pytest.raises(
ValueError,