Goal is to automate building and tuning a classification model to predict employee attrition, using the h2o::h2o.automl.
Import the cleaned data from Module 7.
library(h2o)## Warning: package 'h2o' was built under R version 4.4.3##
## ----------------------------------------------------------------------
##
## Your next step is to start H2O:
## > h2o.init()
##
## For H2O package documentation, ask for help:
## > ??h2o
##
## After starting H2O, you can use the Web UI at http://localhost:54321
## For more information visit https://docs.h2o.ai
##
## ----------------------------------------------------------------------##
## Attaching package: 'h2o'## The following objects are masked from 'package:stats':
##
## cor, sd, var## The following objects are masked from 'package:base':
##
## %*%, %in%, &&, ||, apply, as.factor, as.numeric, colnames,
## colnames<-, ifelse, is.character, is.factor, is.numeric, log,
## log10, log1p, log2, round, signif, trunclibrary(tidyverse)## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.4 ✔ readr 2.1.5
## ✔ forcats 1.0.0 ✔ stringr 1.5.1
## ✔ ggplot2 3.5.1 ✔ tibble 3.2.1
## ✔ lubridate 1.9.3 ✔ tidyr 1.3.1
## ✔ purrr 1.0.2## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ lubridate::day() masks h2o::day()
## ✖ dplyr::filter() masks stats::filter()
## ✖ lubridate::hour() masks h2o::hour()
## ✖ dplyr::lag() masks stats::lag()
## ✖ lubridate::month() masks h2o::month()
## ✖ lubridate::week() masks h2o::week()
## ✖ lubridate::year() masks h2o::year()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorslibrary(tidymodels)## Warning: package 'tidymodels' was built under R version 4.4.2## ── Attaching packages ────────────────────────────────────── tidymodels 1.2.0 ──
## ✔ broom 1.0.6 ✔ rsample 1.2.1
## ✔ dials 1.3.0 ✔ tune 1.2.1
## ✔ infer 1.0.7 ✔ workflows 1.1.4
## ✔ modeldata 1.4.0 ✔ workflowsets 1.1.0
## ✔ parsnip 1.2.1 ✔ yardstick 1.3.2
## ✔ recipes 1.1.0## Warning: package 'dials' was built under R version 4.4.2## Warning: package 'infer' was built under R version 4.4.2## Warning: package 'modeldata' was built under R version 4.4.2## Warning: package 'parsnip' was built under R version 4.4.2## Warning: package 'tune' was built under R version 4.4.2## Warning: package 'workflows' was built under R version 4.4.2## Warning: package 'workflowsets' was built under R version 4.4.3## Warning: package 'yardstick' was built under R version 4.4.2## ── Conflicts ───────────────────────────────────────── tidymodels_conflicts() ──
## ✖ scales::discard() masks purrr::discard()
## ✖ dplyr::filter() masks stats::filter()
## ✖ recipes::fixed() masks stringr::fixed()
## ✖ dplyr::lag() masks stats::lag()
## ✖ yardstick::spec() masks readr::spec()
## ✖ recipes::step() masks stats::step()
## • Learn how to get started at https://www.tidymodels.org/start/library(tidyquant)## Registered S3 method overwritten by 'quantmod':
## method from
## as.zoo.data.frame zoo
## ── Attaching core tidyquant packages ──────────────────────── tidyquant 1.0.9 ──
## ✔ PerformanceAnalytics 2.0.4 ✔ TTR 0.24.4
## ✔ quantmod 0.4.26 ✔ xts 0.14.0── Conflicts ────────────────────────────────────────── tidyquant_conflicts() ──
## ✖ zoo::as.Date() masks base::as.Date()
## ✖ zoo::as.Date.numeric() masks base::as.Date.numeric()
## ✖ scales::col_factor() masks readr::col_factor()
## ✖ lubridate::day() masks h2o::day()
## ✖ scales::discard() masks purrr::discard()
## ✖ dplyr::filter() masks stats::filter()
## ✖ xts::first() masks dplyr::first()
## ✖ recipes::fixed() masks stringr::fixed()
## ✖ lubridate::hour() masks h2o::hour()
## ✖ dplyr::lag() masks stats::lag()
## ✖ xts::last() masks dplyr::last()
## ✖ PerformanceAnalytics::legend() masks graphics::legend()
## ✖ TTR::momentum() masks dials::momentum()
## ✖ lubridate::month() masks h2o::month()
## ✖ yardstick::spec() masks readr::spec()
## ✖ quantmod::summary() masks h2o::summary(), base::summary()
## ✖ lubridate::week() masks h2o::week()
## ✖ lubridate::year() masks h2o::year()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorsdata <- read_csv("../00_data/data_wrangled/data_clean.csv") %>%
# h2o requires all variables to be either numeric or factors
mutate(across(where(is.character), factor))## Rows: 1470 Columns: 32
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (8): Attrition, BusinessTravel, Department, EducationField, Gender, Job...
## dbl (24): Age, DailyRate, DistanceFromHome, Education, EmployeeNumber, Envir...
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.set.seed(1234)
data_split <- initial_split(data, strata = "Attrition")
train_tbl <- training(data_split)
test_tbl <- testing(data_split)recipe_obj <- recipe(Attrition ~ ., data = train_tbl) %>%
# Remove zero variance variables
step_zv(all_predictors()) # Initialize h20
h2o.init()## Connection successful!
##
## R is connected to the H2O cluster:
## H2O cluster uptime: 2 days 22 hours
## H2O cluster timezone: America/New_York
## H2O data parsing timezone: UTC
## H2O cluster version: 3.44.0.3
## H2O cluster version age: 1 year, 4 months and 10 days
## H2O cluster name: H2O_started_from_R_trito_qxv383
## H2O cluster total nodes: 1
## H2O cluster total memory: 3.74 GB
## H2O cluster total cores: 12
## H2O cluster allowed cores: 12
## H2O cluster healthy: TRUE
## H2O Connection ip: localhost
## H2O Connection port: 54321
## H2O Connection proxy: NA
## H2O Internal Security: FALSE
## R Version: R version 4.4.1 (2024-06-14 ucrt)## Warning in h2o.clusterInfo():
## Your H2O cluster version is (1 year, 4 months and 10 days) old. There may be a newer version available.
## Please download and install the latest version from: https://h2o-release.s3.amazonaws.com/h2o/latest_stable.htmlsplit.h2o <- h2o.splitFrame(as.h2o(train_tbl), ratios = c(0.85), seed = 2345)## | | | 0% | |======================================================================| 100%train_h2o <- split.h2o[[1]]
valid_h2o <-split.h2o[[2]]
test_h2o <- as.h2o(test_tbl)## | | | 0% | |======================================================================| 100%y <- "Attrition"
x <- setdiff(names(train_tbl), y)
model_h2o <- h2o.automl(
x = x,
y = y, training_frame = train_h2o,
validation_frame = valid_h2o,
leaderboard_frame = test_h2o,
max_runtime_secs = 30,
nfolds = 5,
seed = 3456
)## | | | 0% | |==== | 6%
## 10:16:38.555: User specified a validation frame with cross-validation still enabled. Please note that the models will still be validated using cross-validation only, the validation frame will be used to provide purely informative validation metrics on the trained models.
## 10:16:38.583: AutoML: XGBoost is not available; skipping it. | |=========== | 16% | |================ | 22% | |====================== | 31% | |============================ | 40% | |=================================== | 49% | |========================================= | 59% | |================================================= | 70% | |======================================================== | 80% | |=============================================================== | 90% | |======================================================================| 100%Examine the output of h2o.automl
model_h2o %>% typeof()## [1] "S4"model_h2o %>% slotNames()## [1] "project_name" "leader" "leaderboard" "event_log"
## [5] "modeling_steps" "training_info"model_h2o@leaderboard## model_id auc logloss
## 1 StackedEnsemble_BestOfFamily_4_AutoML_5_20250501_101638 0.8310680 0.3260493
## 2 StackedEnsemble_BestOfFamily_3_AutoML_5_20250501_101638 0.8288026 0.3246406
## 3 StackedEnsemble_BestOfFamily_2_AutoML_5_20250501_101638 0.8283172 0.3241037
## 4 GLM_1_AutoML_5_20250501_101638 0.8261597 0.3318676
## 5 StackedEnsemble_BestOfFamily_1_AutoML_5_20250501_101638 0.8258900 0.3346208
## 6 StackedEnsemble_AllModels_1_AutoML_5_20250501_101638 0.8235707 0.3281452
## aucpr mean_per_class_error rmse mse
## 1 0.9506023 0.2863269 0.3074067 0.09449890
## 2 0.9508644 0.2677994 0.3069480 0.09421705
## 3 0.9504244 0.2677994 0.3068317 0.09414572
## 4 0.9466326 0.2930421 0.3082111 0.09499409
## 5 0.9494447 0.2627023 0.3115649 0.09707267
## 6 0.9489241 0.3365696 0.3087628 0.09533445
##
## [37 rows x 7 columns]model_h2o@leader## Model Details:
## ==============
##
## H2OBinomialModel: stackedensemble
## Model ID: StackedEnsemble_BestOfFamily_4_AutoML_5_20250501_101638
## Model Summary for Stacked Ensemble:
## key value
## 1 Stacking strategy cross_validation
## 2 Number of base models (used / total) 5/5
## 3 # GBM base models (used / total) 1/1
## 4 # GLM base models (used / total) 1/1
## 5 # DeepLearning base models (used / total) 1/1
## 6 # DRF base models (used / total) 2/2
## 7 Metalearner algorithm GLM
## 8 Metalearner fold assignment scheme Random
## 9 Metalearner nfolds 5
## 10 Metalearner fold_column NA
## 11 Custom metalearner hyperparameters None
##
##
## H2OBinomialMetrics: stackedensemble
## ** Reported on training data. **
##
## MSE: 0.06772075
## RMSE: 0.2602321
## LogLoss: 0.23784
## Mean Per-Class Error: 0.1640377
## AUC: 0.9331792
## AUCPR: 0.9823192
## Gini: 0.8663583
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 109 49 0.310127 =49/158
## No 14 766 0.017949 =14/780
## Totals 123 815 0.067164 =63/938
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.580770 0.960502 293
## 2 max f2 0.546634 0.977463 304
## 3 max f0point5 0.748990 0.950855 235
## 4 max accuracy 0.593085 0.932836 290
## 5 max precision 0.999038 1.000000 0
## 6 max recall 0.313609 1.000000 364
## 7 max specificity 0.999038 1.000000 0
## 8 max absolute_mcc 0.593085 0.746036 290
## 9 max min_per_class_accuracy 0.801723 0.867089 207
## 10 max mean_per_class_accuracy 0.801723 0.867519 207
## 11 max tns 0.999038 158.000000 0
## 12 max fns 0.999038 774.000000 0
## 13 max fps 0.043694 158.000000 399
## 14 max tps 0.313609 780.000000 364
## 15 max tnr 0.999038 1.000000 0
## 16 max fnr 0.999038 0.992308 0
## 17 max fpr 0.043694 1.000000 399
## 18 max tpr 0.313609 1.000000 364
##
## Gains/Lift Table: Extract with `h2o.gainsLift(<model>, <data>)` or `h2o.gainsLift(<model>, valid=<T/F>, xval=<T/F>)`
## H2OBinomialMetrics: stackedensemble
## ** Reported on validation data. **
##
## MSE: 0.08380031
## RMSE: 0.2894828
## LogLoss: 0.3065863
## Mean Per-Class Error: 0.3684211
## AUC: 0.745614
## AUCPR: 0.9468608
## Gini: 0.4912281
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 5 14 0.736842 =14/19
## No 0 144 0.000000 =0/144
## Totals 5 158 0.085890 =14/163
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.306706 0.953642 157
## 2 max f2 0.306706 0.980926 157
## 3 max f0point5 0.578313 0.940860 149
## 4 max accuracy 0.578313 0.914110 149
## 5 max precision 0.999399 1.000000 0
## 6 max recall 0.306706 1.000000 157
## 7 max specificity 0.999399 1.000000 0
## 8 max absolute_mcc 0.578313 0.528183 149
## 9 max min_per_class_accuracy 0.874376 0.652778 99
## 10 max mean_per_class_accuracy 0.578313 0.722953 149
## 11 max tns 0.999399 19.000000 0
## 12 max fns 0.999399 143.000000 0
## 13 max fps 0.089616 19.000000 162
## 14 max tps 0.306706 144.000000 157
## 15 max tnr 0.999399 1.000000 0
## 16 max fnr 0.999399 0.993056 0
## 17 max fpr 0.089616 1.000000 162
## 18 max tpr 0.306706 1.000000 157
##
## Gains/Lift Table: Extract with `h2o.gainsLift(<model>, <data>)` or `h2o.gainsLift(<model>, valid=<T/F>, xval=<T/F>)`
## H2OBinomialMetrics: stackedensemble
## ** Reported on cross-validation data. **
## ** 5-fold cross-validation on training data (Metrics computed for combined holdout predictions) **
##
## MSE: 0.09579919
## RMSE: 0.3095144
## LogLoss: 0.3350831
## Mean Per-Class Error: 0.3122119
## AUC: 0.8413908
## AUCPR: 0.9450908
## Gini: 0.6827816
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 64 94 0.594937 =94/158
## No 23 757 0.029487 =23/780
## Totals 87 851 0.124733 =117/938
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.540214 0.928265 324
## 2 max f2 0.369443 0.965414 366
## 3 max f0point5 0.750401 0.920405 231
## 4 max accuracy 0.540214 0.875267 324
## 5 max precision 0.967579 0.963801 48
## 6 max recall 0.257446 1.000000 385
## 7 max specificity 0.999919 0.993671 0
## 8 max absolute_mcc 0.638493 0.519326 289
## 9 max min_per_class_accuracy 0.826052 0.766667 185
## 10 max mean_per_class_accuracy 0.783037 0.790725 214
## 11 max tns 0.999919 157.000000 0
## 12 max fns 0.999919 768.000000 0
## 13 max fps 0.066296 158.000000 399
## 14 max tps 0.257446 780.000000 385
## 15 max tnr 0.999919 0.993671 0
## 16 max fnr 0.999919 0.984615 0
## 17 max fpr 0.066296 1.000000 399
## 18 max tpr 0.257446 1.000000 385
##
## Gains/Lift Table: Extract with `h2o.gainsLift(<model>, <data>)` or `h2o.gainsLift(<model>, valid=<T/F>, xval=<T/F>)`
## Cross-Validation Metrics Summary:
## mean sd cv_1_valid cv_2_valid cv_3_valid cv_4_valid
## accuracy 0.889566 0.015523 0.895238 0.880829 0.871508 0.887755
## auc 0.835847 0.040489 0.799771 0.869767 0.888691 0.814278
## err 0.110434 0.015523 0.104762 0.119171 0.128492 0.112245
## err_count 20.800000 3.834058 22.000000 23.000000 23.000000 22.000000
## f0point5 0.914734 0.009304 0.920916 0.912596 0.903141 0.910125
## cv_5_valid
## accuracy 0.912500
## auc 0.806729
## err 0.087500
## err_count 14.000000
## f0point5 0.926893
##
## ---
## mean sd cv_1_valid cv_2_valid cv_3_valid
## precision 0.901317 0.008761 0.907692 0.904459 0.890323
## r2 0.296878 0.081217 0.234515 0.380630 0.352591
## recall 0.972899 0.020829 0.977901 0.946667 0.958333
## residual_deviance 125.231690 19.540842 135.046830 140.039440 124.439480
## rmse 0.307976 0.016568 0.301847 0.327490 0.319118
## specificity 0.438245 0.159257 0.379310 0.651163 0.514286
## cv_4_valid cv_5_valid
## precision 0.893855 0.910256
## r2 0.326900 0.189753
## recall 0.981595 1.000000
## residual_deviance 134.847640 91.785060
## rmse 0.306997 0.284426
## specificity 0.424242 0.222222?h2o.getModel## starting httpd help server ... done?h2o.saveModel
?h2o.loadModel
# h2o.getModel("GLM_1_AutoML_1_20250428_124032") %>%
# h2o.saveModel("h2o_models/")
#
# best_model <- h2o.loadModel("h2o_models/GLM_1_AutoML_1_20250428_124032")
best_model <- model_h2o@leaderpredictions <- h2o.predict(best_model, newdata= test_h2o)## | | | 0% | |======================================================================| 100%predictions_tibble <- predictions %>%
as_tibble()
predictions_tibble %>%
bind_cols(test_tbl)## # A tibble: 369 × 35
## predict Left No Age Attrition BusinessTravel DailyRate Department
## <fct> <dbl> <dbl> <dbl> <fct> <fct> <dbl> <fct>
## 1 No 0.533 0.467 41 Left Travel_Rarely 1102 Sales
## 2 No 0.0184 0.982 49 No Travel_Frequently 279 Research & …
## 3 No 0.256 0.744 33 No Travel_Frequently 1392 Research & …
## 4 No 0.185 0.815 59 No Travel_Rarely 1324 Research & …
## 5 No 0.0588 0.941 38 No Travel_Frequently 216 Research & …
## 6 No 0.298 0.702 29 No Travel_Rarely 153 Research & …
## 7 No 0.0635 0.937 34 No Travel_Rarely 1346 Research & …
## 8 Left 0.847 0.153 28 Left Travel_Rarely 103 Research & …
## 9 No 0.341 0.659 22 No Non-Travel 1123 Research & …
## 10 No 0.0203 0.980 53 No Travel_Rarely 1219 Sales
## # ℹ 359 more rows
## # ℹ 27 more variables: DistanceFromHome <dbl>, Education <dbl>,
## # EducationField <fct>, EmployeeNumber <dbl>, EnvironmentSatisfaction <dbl>,
## # Gender <fct>, HourlyRate <dbl>, JobInvolvement <dbl>, JobLevel <dbl>,
## # JobRole <fct>, JobSatisfaction <dbl>, MaritalStatus <fct>,
## # MonthlyIncome <dbl>, MonthlyRate <dbl>, NumCompaniesWorked <dbl>,
## # OverTime <fct>, PercentSalaryHike <dbl>, PerformanceRating <dbl>, …?h2o.performance
performance_h2o <- h2o.performance(best_model, newdata = test_h2o)
typeof(performance_h2o)## [1] "S4"slotNames(performance_h2o)## [1] "algorithm" "on_train" "on_valid" "on_xval" "metrics"performance_h2o@metrics## $model
## $model$`__meta`
## $model$`__meta`$schema_version
## [1] 3
##
## $model$`__meta`$schema_name
## [1] "ModelKeyV3"
##
## $model$`__meta`$schema_type
## [1] "Key<Model>"
##
##
## $model$name
## [1] "StackedEnsemble_BestOfFamily_4_AutoML_5_20250501_101638"
##
## $model$type
## [1] "Key<Model>"
##
## $model$URL
## [1] "/3/Models/StackedEnsemble_BestOfFamily_4_AutoML_5_20250501_101638"
##
##
## $model_checksum
## [1] "1404117915454927488"
##
## $frame
## $frame$name
## [1] "test_tbl_sid_a88a_3"
##
##
## $frame_checksum
## [1] "-54192601206779456"
##
## $description
## NULL
##
## $scoring_time
## [1] 1.746109e+12
##
## $predictions
## NULL
##
## $MSE
## [1] 0.0944989
##
## $RMSE
## [1] 0.3074067
##
## $nobs
## [1] 369
##
## $custom_metric_name
## NULL
##
## $custom_metric_value
## [1] 0
##
## $r2
## [1] 0.3059836
##
## $logloss
## [1] 0.3260493
##
## $AUC
## [1] 0.831068
##
## $pr_auc
## [1] 0.9506023
##
## $Gini
## [1] 0.6621359
##
## $mean_per_class_error
## [1] 0.2863269
##
## $domain
## [1] "Left" "No"
##
## $cm
## $cm$`__meta`
## $cm$`__meta`$schema_version
## [1] 3
##
## $cm$`__meta`$schema_name
## [1] "ConfusionMatrixV3"
##
## $cm$`__meta`$schema_type
## [1] "ConfusionMatrix"
##
##
## $cm$table
## Confusion Matrix: Row labels: Actual class; Column labels: Predicted class
## Left No Error Rate
## Left 27 33 0.5500 = 33 / 60
## No 7 302 0.0227 = 7 / 309
## Totals 34 335 0.1084 = 40 / 369
##
##
## $thresholds_and_metric_scores
## Metrics for Thresholds: Binomial metrics as a function of classification thresholds
## threshold f1 f2 f0point5 accuracy precision recall specificity
## 1 0.999026 0.006452 0.004042 0.015974 0.165312 1.000000 0.003236 1.000000
## 2 0.998845 0.012862 0.008078 0.031546 0.168022 1.000000 0.006472 1.000000
## 3 0.998231 0.019231 0.012107 0.046729 0.170732 1.000000 0.009709 1.000000
## 4 0.998207 0.025559 0.016129 0.061538 0.173442 1.000000 0.012945 1.000000
## 5 0.998061 0.031847 0.020145 0.075988 0.176152 1.000000 0.016181 1.000000
## absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns fns fps tps
## 1 0.022971 0.003236 0.501618 60 308 0 1
## 2 0.032530 0.006472 0.503236 60 307 0 2
## 3 0.039895 0.009709 0.504854 60 306 0 3
## 4 0.046130 0.012945 0.506472 60 305 0 4
## 5 0.051645 0.016181 0.508091 60 304 0 5
## tnr fnr fpr tpr idx
## 1 1.000000 0.996764 0.000000 0.003236 0
## 2 1.000000 0.993528 0.000000 0.006472 1
## 3 1.000000 0.990291 0.000000 0.009709 2
## 4 1.000000 0.987055 0.000000 0.012945 3
## 5 1.000000 0.983819 0.000000 0.016181 4
##
## ---
## threshold f1 f2 f0point5 accuracy precision recall
## 364 0.245303 0.918276 0.965625 0.875354 0.850949 0.848901 1.000000
## 365 0.238439 0.916914 0.965022 0.873375 0.848238 0.846575 1.000000
## 366 0.167989 0.915556 0.964419 0.871404 0.845528 0.844262 1.000000
## 367 0.153099 0.914201 0.963818 0.869443 0.842818 0.841962 1.000000
## 368 0.092234 0.912851 0.963217 0.867490 0.840108 0.839674 1.000000
## 369 0.071644 0.911504 0.962617 0.865546 0.837398 0.837398 1.000000
## specificity absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns
## 364 0.083333 0.265973 0.083333 0.541667 5
## 365 0.066667 0.237568 0.066667 0.533333 4
## 366 0.050000 0.205458 0.050000 0.525000 3
## 367 0.033333 0.167527 0.033333 0.516667 2
## 368 0.016667 0.118299 0.016667 0.508333 1
## 369 0.000000 0.000000 0.000000 0.500000 0
## fns fps tps tnr fnr fpr tpr idx
## 364 0 55 309 0.083333 0.000000 0.916667 1.000000 363
## 365 0 56 309 0.066667 0.000000 0.933333 1.000000 364
## 366 0 57 309 0.050000 0.000000 0.950000 1.000000 365
## 367 0 58 309 0.033333 0.000000 0.966667 1.000000 366
## 368 0 59 309 0.016667 0.000000 0.983333 1.000000 367
## 369 0 60 309 0.000000 0.000000 1.000000 1.000000 368
##
## $max_criteria_and_metric_scores
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.496084 0.937888 334
## 2 max f2 0.272203 0.966834 361
## 3 max f0point5 0.573626 0.920645 325
## 4 max accuracy 0.496084 0.891599 334
## 5 max precision 0.999026 1.000000 0
## 6 max recall 0.272203 1.000000 361
## 7 max specificity 0.999026 1.000000 0
## 8 max absolute_mcc 0.573626 0.549524 325
## 9 max min_per_class_accuracy 0.838000 0.760518 246
## 10 max mean_per_class_accuracy 0.838000 0.780259 246
## 11 max tns 0.999026 60.000000 0
## 12 max fns 0.999026 308.000000 0
## 13 max fps 0.071644 60.000000 368
## 14 max tps 0.272203 309.000000 361
## 15 max tnr 0.999026 1.000000 0
## 16 max fnr 0.999026 0.996764 0
## 17 max fpr 0.071644 1.000000 368
## 18 max tpr 0.272203 1.000000 361
##
## $gains_lift_table
## Gains/Lift Table: Avg response rate: 83.74 %, avg score: 83.52 %
## group cumulative_data_fraction lower_threshold lift cumulative_lift
## 1 1 0.01084011 0.998108 1.194175 1.194175
## 2 2 0.02168022 0.997224 1.194175 1.194175
## 3 3 0.03252033 0.996300 1.194175 1.194175
## 4 4 0.04065041 0.995619 1.194175 1.194175
## 5 5 0.05149051 0.994995 1.194175 1.194175
## 6 6 0.10027100 0.989375 1.061489 1.129625
## 7 7 0.15176152 0.984816 1.194175 1.151526
## 8 8 0.20054201 0.976642 1.127832 1.145762
## 9 9 0.30081301 0.961479 1.129625 1.140383
## 10 10 0.40108401 0.943573 1.194175 1.153831
## 11 11 0.50135501 0.913399 1.065075 1.136080
## 12 12 0.59891599 0.871039 1.161003 1.140140
## 13 13 0.69918699 0.819204 1.000525 1.120117
## 14 14 0.79945799 0.723243 0.935975 1.097022
## 15 15 0.89972900 0.520020 0.935975 1.079074
## 16 16 1.00000000 0.071644 0.290475 1.000000
## response_rate score cumulative_response_rate cumulative_score
## 1 1.000000 0.998577 1.000000 0.998577
## 2 1.000000 0.997733 1.000000 0.998155
## 3 1.000000 0.996737 1.000000 0.997682
## 4 1.000000 0.995973 1.000000 0.997341
## 5 1.000000 0.995359 1.000000 0.996923
## 6 0.888889 0.991732 0.945946 0.994398
## 7 1.000000 0.987278 0.964286 0.991982
## 8 0.944444 0.979428 0.959459 0.988929
## 9 0.945946 0.970265 0.954955 0.982707
## 10 1.000000 0.952835 0.966216 0.975239
## 11 0.891892 0.931661 0.951351 0.966523
## 12 0.972222 0.894858 0.954751 0.954849
## 13 0.837838 0.847173 0.937984 0.939407
## 14 0.783784 0.779751 0.918644 0.919383
## 15 0.783784 0.643493 0.903614 0.888636
## 16 0.243243 0.355589 0.837398 0.835187
## capture_rate cumulative_capture_rate gain cumulative_gain
## 1 0.012945 0.012945 19.417476 19.417476
## 2 0.012945 0.025890 19.417476 19.417476
## 3 0.012945 0.038835 19.417476 19.417476
## 4 0.009709 0.048544 19.417476 19.417476
## 5 0.012945 0.061489 19.417476 19.417476
## 6 0.051780 0.113269 6.148867 12.962477
## 7 0.061489 0.174757 19.417476 15.152566
## 8 0.055016 0.229773 12.783172 14.576227
## 9 0.113269 0.343042 12.962477 14.038310
## 10 0.119741 0.462783 19.417476 15.383102
## 11 0.106796 0.569579 6.507478 13.607977
## 12 0.113269 0.682848 16.100324 14.013970
## 13 0.100324 0.783172 0.052480 12.011741
## 14 0.093851 0.877023 -6.402519 9.702156
## 15 0.093851 0.970874 -6.402519 7.907358
## 16 0.029126 1.000000 -70.952506 0.000000
## kolmogorov_smirnov
## 1 0.012945
## 2 0.025890
## 3 0.038835
## 4 0.048544
## 5 0.061489
## 6 0.079935
## 7 0.141424
## 8 0.179773
## 9 0.259709
## 10 0.379450
## 11 0.419579
## 12 0.516181
## 13 0.516505
## 14 0.477023
## 15 0.437540
## 16 0.000000
##
## $residual_deviance
## [1] 240.6244
##
## $null_deviance
## [1] 327.7324
##
## $AIC
## [1] 252.6244
##
## $loglikelihood
## [1] 0
##
## $null_degrees_of_freedom
## [1] 368
##
## $residual_degrees_of_freedom
## [1] 363h2o.auc(performance_h2o)## [1] 0.831068h2o.confusionMatrix(performance_h2o)## Confusion Matrix (vertical: actual; across: predicted) for max f1 @ threshold = 0.496084097876897:
## Left No Error Rate
## Left 27 33 0.550000 =33/60
## No 7 302 0.022654 =7/309
## Totals 34 335 0.108401 =40/369h2o.metric(performance_h2o)## Metrics for Thresholds: Binomial metrics as a function of classification thresholds
## threshold f1 f2 f0point5 accuracy precision recall specificity
## 1 0.999026 0.006452 0.004042 0.015974 0.165312 1.000000 0.003236 1.000000
## 2 0.998845 0.012862 0.008078 0.031546 0.168022 1.000000 0.006472 1.000000
## 3 0.998231 0.019231 0.012107 0.046729 0.170732 1.000000 0.009709 1.000000
## 4 0.998207 0.025559 0.016129 0.061538 0.173442 1.000000 0.012945 1.000000
## 5 0.998061 0.031847 0.020145 0.075988 0.176152 1.000000 0.016181 1.000000
## absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns fns fps tps
## 1 0.022971 0.003236 0.501618 60 308 0 1
## 2 0.032530 0.006472 0.503236 60 307 0 2
## 3 0.039895 0.009709 0.504854 60 306 0 3
## 4 0.046130 0.012945 0.506472 60 305 0 4
## 5 0.051645 0.016181 0.508091 60 304 0 5
## tnr fnr fpr tpr idx
## 1 1.000000 0.996764 0.000000 0.003236 0
## 2 1.000000 0.993528 0.000000 0.006472 1
## 3 1.000000 0.990291 0.000000 0.009709 2
## 4 1.000000 0.987055 0.000000 0.012945 3
## 5 1.000000 0.983819 0.000000 0.016181 4
##
## ---
## threshold f1 f2 f0point5 accuracy precision recall
## 364 0.245303 0.918276 0.965625 0.875354 0.850949 0.848901 1.000000
## 365 0.238439 0.916914 0.965022 0.873375 0.848238 0.846575 1.000000
## 366 0.167989 0.915556 0.964419 0.871404 0.845528 0.844262 1.000000
## 367 0.153099 0.914201 0.963818 0.869443 0.842818 0.841962 1.000000
## 368 0.092234 0.912851 0.963217 0.867490 0.840108 0.839674 1.000000
## 369 0.071644 0.911504 0.962617 0.865546 0.837398 0.837398 1.000000
## specificity absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns
## 364 0.083333 0.265973 0.083333 0.541667 5
## 365 0.066667 0.237568 0.066667 0.533333 4
## 366 0.050000 0.205458 0.050000 0.525000 3
## 367 0.033333 0.167527 0.033333 0.516667 2
## 368 0.016667 0.118299 0.016667 0.508333 1
## 369 0.000000 0.000000 0.000000 0.500000 0
## fns fps tps tnr fnr fpr tpr idx
## 364 0 55 309 0.083333 0.000000 0.916667 1.000000 363
## 365 0 56 309 0.066667 0.000000 0.933333 1.000000 364
## 366 0 57 309 0.050000 0.000000 0.950000 1.000000 365
## 367 0 58 309 0.033333 0.000000 0.966667 1.000000 366
## 368 0 59 309 0.016667 0.000000 0.983333 1.000000 367
## 369 0 60 309 0.000000 0.000000 1.000000 1.000000 368