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)##
## ----------------------------------------------------------------------
##
## 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)## ── Attaching packages ────────────────────────────────────── tidymodels 1.2.0 ──
## ✔ broom 1.0.8 ✔ 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
## ── 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()
## • Search for functions across packages at https://www.tidymodels.org/find/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 h2o
h2o.init()## Connection successful!
##
## R is connected to the H2O cluster:
## H2O cluster uptime: 1 days 1 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 3 days
## H2O cluster name: H2O_started_from_R_alyssadalessio_fyb567
## H2O cluster total nodes: 1
## H2O cluster total memory: 3.23 GB
## H2O cluster total cores: 8
## H2O cluster allowed cores: 8
## 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)## Warning in h2o.clusterInfo():
## Your H2O cluster version is (1 year, 4 months and 3 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)
models_h2o <- h2o.automl(
x = x,
y = y,
training_frame = train_h2o,
validation_frame = valid_h2o,
leaderboard_frame = test_h2o,
# max_runtime_secs = 30,
max_models = 10,
exclude_algos = "DeepLearning",
nfolds = 5,
seed = 3456
)## | | | 0% | |==== | 5%
## 20:12:02.727: 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. | |============= | 18% | |====================== | 31% | |======================================================================| 100%Evaluate the output of h2o.automl
models_h2o %>% typeof()## [1] "S4"models_h2o %>% slotNames()## [1] "project_name" "leader" "leaderboard" "event_log"
## [5] "modeling_steps" "training_info"models_h2o@leaderboard## model_id auc logloss
## 1 StackedEnsemble_BestOfFamily_1_AutoML_11_20250423_201202 0.8336570 0.3250877
## 2 StackedEnsemble_AllModels_1_AutoML_11_20250423_201202 0.8275620 0.3311942
## 3 GLM_1_AutoML_11_20250423_201202 0.8261597 0.3318676
## 4 GBM_2_AutoML_11_20250423_201202 0.8080906 0.3494185
## 5 GBM_1_AutoML_11_20250423_201202 0.8071197 0.3487110
## 6 XGBoost_1_AutoML_11_20250423_201202 0.8033981 0.3540796
## aucpr mean_per_class_error rmse mse
## 1 0.9524284 0.3231392 0.3086336 0.09525469
## 2 0.9511222 0.2895631 0.3124007 0.09759422
## 3 0.9466326 0.2930421 0.3082111 0.09499409
## 4 0.9476344 0.3379450 0.3243230 0.10518538
## 5 0.9495705 0.3177994 0.3248711 0.10554125
## 6 0.9408309 0.3513754 0.3235331 0.10467365
##
## [12 rows x 7 columns]models_h2o@leader## Model Details:
## ==============
##
## H2OBinomialModel: stackedensemble
## Model ID: StackedEnsemble_BestOfFamily_1_AutoML_11_20250423_201202
## Model Summary for Stacked Ensemble:
## key value
## 1 Stacking strategy cross_validation
## 2 Number of base models (used / total) 4/5
## 3 # GBM base models (used / total) 1/1
## 4 # XGBoost base models (used / total) 1/1
## 5 # GLM base models (used / total) 1/1
## 6 # DRF base models (used / total) 1/2
## 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: 0.04039314
## RMSE: 0.2009804
## LogLoss: 0.1597266
## Mean Per-Class Error: 0.07919507
## AUC: 0.979828
## AUCPR: 0.9952117
## Gini: 0.959656
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 135 23 0.145570 =23/158
## No 10 770 0.012821 =10/780
## Totals 145 793 0.035181 =33/938
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.614781 0.979021 271
## 2 max f2 0.406882 0.987842 305
## 3 max f0point5 0.679826 0.978822 254
## 4 max accuracy 0.657200 0.964819 262
## 5 max precision 0.999563 1.000000 0
## 6 max recall 0.406882 1.000000 305
## 7 max specificity 0.999563 1.000000 0
## 8 max absolute_mcc 0.657200 0.873516 262
## 9 max min_per_class_accuracy 0.763278 0.930380 227
## 10 max mean_per_class_accuracy 0.740545 0.939549 237
## 11 max tns 0.999563 158.000000 0
## 12 max fns 0.999563 778.000000 0
## 13 max fps 0.021597 158.000000 399
## 14 max tps 0.406882 780.000000 305
## 15 max tnr 0.999563 1.000000 0
## 16 max fnr 0.999563 0.997436 0
## 17 max fpr 0.021597 1.000000 399
## 18 max tpr 0.406882 1.000000 305
##
## 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: 0.08564897
## RMSE: 0.2926585
## LogLoss: 0.3030019
## Mean Per-Class Error: 0.3455775
## AUC: 0.7627924
## AUCPR: 0.9562414
## Gini: 0.5255848
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 6 13 0.684211 =13/19
## No 1 143 0.006944 =1/144
## Totals 7 156 0.085890 =14/163
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.385576 0.953333 155
## 2 max f2 0.203388 0.979592 158
## 3 max f0point5 0.453411 0.934211 153
## 4 max accuracy 0.453411 0.914110 153
## 5 max precision 0.999475 1.000000 0
## 6 max recall 0.203388 1.000000 158
## 7 max specificity 0.999475 1.000000 0
## 8 max absolute_mcc 0.453411 0.498118 153
## 9 max min_per_class_accuracy 0.886259 0.659722 100
## 10 max mean_per_class_accuracy 0.646522 0.705592 144
## 11 max tns 0.999475 19.000000 0
## 12 max fns 0.999475 143.000000 0
## 13 max fps 0.047843 19.000000 162
## 14 max tps 0.203388 144.000000 158
## 15 max tnr 0.999475 1.000000 0
## 16 max fnr 0.999475 0.993056 0
## 17 max fpr 0.047843 1.000000 162
## 18 max tpr 0.203388 1.000000 158
##
## 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: 0.09449772
## RMSE: 0.3074048
## LogLoss: 0.3212977
## Mean Per-Class Error: 0.3001947
## AUC: 0.8490506
## AUCPR: 0.9547088
## Gini: 0.6981013
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 68 90 0.569620 =90/158
## No 24 756 0.030769 =24/780
## Totals 92 846 0.121535 =114/938
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.513265 0.929889 317
## 2 max f2 0.184838 0.964153 387
## 3 max f0point5 0.718349 0.922810 241
## 4 max accuracy 0.515316 0.878465 316
## 5 max precision 0.999181 1.000000 0
## 6 max recall 0.184838 1.000000 387
## 7 max specificity 0.999181 1.000000 0
## 8 max absolute_mcc 0.636646 0.536015 279
## 9 max min_per_class_accuracy 0.841613 0.772152 181
## 10 max mean_per_class_accuracy 0.790133 0.790084 209
## 11 max tns 0.999181 158.000000 0
## 12 max fns 0.999181 778.000000 0
## 13 max fps 0.017583 158.000000 399
## 14 max tps 0.184838 780.000000 387
## 15 max tnr 0.999181 1.000000 0
## 16 max fnr 0.999181 0.997436 0
## 17 max fpr 0.017583 1.000000 399
## 18 max tpr 0.184838 1.000000 387
##
## 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 0.891432 0.024434 0.909091 0.886010 0.855556 0.887755
## auc 0.849945 0.026559 0.862076 0.889922 0.838818 0.836959
## err 0.108568 0.024434 0.090909 0.113990 0.144444 0.112245
## err_count 20.400000 4.827007 19.000000 22.000000 26.000000 22.000000
## f0point5 0.916626 0.025622 0.935175 0.900243 0.881988 0.920897
## cv_5_valid
## accuracy 0.918750
## auc 0.821948
## err 0.081250
## err_count 13.000000
## f0point5 0.944828
##
## ---
## mean sd cv_1_valid cv_2_valid cv_3_valid
## precision 0.903702 0.032252 0.926316 0.880952 0.860606
## r2 0.311269 0.052565 0.345603 0.373210 0.290453
## recall 0.973408 0.010980 0.972376 0.986667 0.979310
## residual_deviance 120.180690 17.574808 113.260960 134.430590 130.997650
## rmse 0.306345 0.026419 0.275545 0.329445 0.333377
## specificity 0.489902 0.083904 0.500000 0.534884 0.342857
## cv_4_valid cv_5_valid
## precision 0.912281 0.938356
## r2 0.310664 0.236416
## recall 0.957055 0.971631
## residual_deviance 129.853960 92.360300
## rmse 0.310678 0.282680
## specificity 0.545455 0.526316?h2o.getModel
?h2o.saveModel
?h2o.loadModel
# h2o.getModel("GLM_1_AutoML_6_20250422_204739") %>%
# h2o.saveModel("h2o_models/")
# h2o.getModel("StackedEnsemble_BestOfFamily_1_AutoML_6_20250422_204739") %>%
# h2o.saveModel("h2o_models/")
best_model <- models_h2o@leaderpredictions <- h2o.predict(best_model, newdata = test_h2o)## | | | 0% | |======================================================================| 100%predictions_tbl <- predictions %>%
as_tibble()
predictions_tbl %>%
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 0.571 0.429 41 Left Travel_Rarely 1102 Sales
## 2 No 0.0186 0.981 49 No Travel_Frequently 279 Research &…
## 3 No 0.410 0.590 33 No Travel_Frequently 1392 Research &…
## 4 No 0.240 0.760 59 No Travel_Rarely 1324 Research &…
## 5 No 0.0455 0.954 38 No Travel_Frequently 216 Research &…
## 6 No 0.341 0.659 29 No Travel_Rarely 153 Research &…
## 7 No 0.0342 0.966 34 No Travel_Rarely 1346 Research &…
## 8 Left 0.920 0.0804 28 Left Travel_Rarely 103 Research &…
## 9 No 0.368 0.632 22 No Non-Travel 1123 Research &…
## 10 No 0.0268 0.973 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_1_AutoML_11_20250423_201202"
##
## $model$type
## [1] "Key<Model>"
##
## $model$URL
## [1] "/3/Models/StackedEnsemble_BestOfFamily_1_AutoML_11_20250423_201202"
##
##
## $model_checksum
## [1] "-1353109486293577944"
##
## $frame
## $frame$name
## [1] "test_tbl_sid_84df_3"
##
##
## $frame_checksum
## [1] "-54192601206779456"
##
## $description
## NULL
##
## $scoring_time
## [1] 1.745454e+12
##
## $predictions
## NULL
##
## $MSE
## [1] 0.09525469
##
## $RMSE
## [1] 0.3086336
##
## $nobs
## [1] 369
##
## $custom_metric_name
## NULL
##
## $custom_metric_value
## [1] 0
##
## $r2
## [1] 0.3004329
##
## $logloss
## [1] 0.3250877
##
## $AUC
## [1] 0.833657
##
## $pr_auc
## [1] 0.9524284
##
## $Gini
## [1] 0.6673139
##
## $mean_per_class_error
## [1] 0.3231392
##
## $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 22 38 0.6333 = 38 / 60
## No 4 305 0.0129 = 4 / 309
## Totals 26 343 0.1138 = 42 / 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 0.999138 0.006452 0.004042 0.015974 0.165312 1.000000 0.003236 1.000000
## 2 0.998393 0.012862 0.008078 0.031546 0.168022 1.000000 0.006472 1.000000
## 3 0.998301 0.019231 0.012107 0.046729 0.170732 1.000000 0.009709 1.000000
## 4 0.998284 0.025559 0.016129 0.061538 0.173442 1.000000 0.012945 1.000000
## 5 0.998197 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 0.186603 0.918276 0.965625 0.875354 0.850949 0.848901 1.000000
## 365 0.184337 0.916914 0.965022 0.873375 0.848238 0.846575 1.000000
## 366 0.093343 0.915556 0.964419 0.871404 0.845528 0.844262 1.000000
## 367 0.080355 0.914201 0.963818 0.869443 0.842818 0.841962 1.000000
## 368 0.063383 0.912851 0.963217 0.867490 0.840108 0.839674 1.000000
## 369 0.040375 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 0.338453 0.935583 342
## 2 max f2 0.331377 0.967130 345
## 3 max f0point5 0.803109 0.916003 267
## 4 max accuracy 0.457557 0.886179 332
## 5 max precision 0.999138 1.000000 0
## 6 max recall 0.186603 1.000000 363
## 7 max specificity 0.999138 1.000000 0
## 8 max absolute_mcc 0.457557 0.523400 332
## 9 max min_per_class_accuracy 0.830093 0.770227 250
## 10 max mean_per_class_accuracy 0.803109 0.784385 267
## 11 max tns 0.999138 60.000000 0
## 12 max fns 0.999138 308.000000 0
## 13 max fps 0.040375 60.000000 368
## 14 max tps 0.186603 309.000000 363
## 15 max tnr 0.999138 1.000000 0
## 16 max fnr 0.999138 0.996764 0
## 17 max fpr 0.040375 1.000000 368
## 18 max tpr 0.186603 1.000000 363
##
## $gains_lift_table
## Gains/Lift Table: Avg response rate: 83.74 %, avg score: 82.82 %
## group cumulative_data_fraction lower_threshold lift cumulative_lift
## 1 1 0.01084011 0.998225 1.194175 1.194175
## 2 2 0.02168022 0.997247 1.194175 1.194175
## 3 3 0.03252033 0.996829 1.194175 1.194175
## 4 4 0.04065041 0.995987 1.194175 1.194175
## 5 5 0.05149051 0.994895 1.194175 1.194175
## 6 6 0.10027100 0.990247 1.127832 1.161900
## 7 7 0.15176152 0.983997 1.131323 1.151526
## 8 8 0.20054201 0.979805 1.127832 1.145762
## 9 9 0.30081301 0.966587 1.129625 1.140383
## 10 10 0.40108401 0.947058 1.161900 1.145762
## 11 11 0.50135501 0.922632 1.129625 1.142535
## 12 12 0.59891599 0.883244 1.161003 1.145543
## 13 13 0.69918699 0.817651 1.032800 1.129375
## 14 14 0.79945799 0.702771 0.871425 1.097022
## 15 15 0.89972900 0.501538 0.903700 1.075477
## 16 16 1.00000000 0.040375 0.322750 1.000000
## response_rate score cumulative_response_rate cumulative_score
## 1 1.000000 0.998529 1.000000 0.998529
## 2 1.000000 0.997770 1.000000 0.998149
## 3 1.000000 0.996975 1.000000 0.997758
## 4 1.000000 0.996245 1.000000 0.997455
## 5 1.000000 0.995628 1.000000 0.997071
## 6 0.944444 0.992047 0.972973 0.994627
## 7 0.947368 0.987033 0.964286 0.992050
## 8 0.944444 0.982008 0.959459 0.989608
## 9 0.945946 0.973794 0.954955 0.984336
## 10 0.972973 0.957572 0.959459 0.977645
## 11 0.945946 0.935275 0.956757 0.969171
## 12 0.972222 0.903677 0.959276 0.958503
## 13 0.864865 0.854414 0.945736 0.943575
## 14 0.729730 0.772793 0.918644 0.922155
## 15 0.756757 0.624763 0.900602 0.889012
## 16 0.270270 0.282454 0.837398 0.828192
## 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 0.055016 0.116505 12.783172 16.189976
## 7 0.058252 0.174757 13.132345 15.152566
## 8 0.055016 0.229773 12.783172 14.576227
## 9 0.113269 0.343042 12.962477 14.038310
## 10 0.116505 0.459547 16.189976 14.576227
## 11 0.113269 0.572816 12.962477 14.253477
## 12 0.113269 0.686084 16.100324 14.554321
## 13 0.103560 0.789644 3.279979 12.937458
## 14 0.087379 0.877023 -12.857518 9.702156
## 15 0.090615 0.967638 -9.630018 7.547666
## 16 0.032362 1.000000 -67.725007 0.000000
## kolmogorov_smirnov
## 1 0.012945
## 2 0.025890
## 3 0.038835
## 4 0.048544
## 5 0.061489
## 6 0.099838
## 7 0.141424
## 8 0.179773
## 9 0.259709
## 10 0.359547
## 11 0.439482
## 12 0.536084
## 13 0.556311
## 14 0.477023
## 15 0.417638
## 16 0.000000
##
## $residual_deviance
## [1] 239.9147
##
## $null_deviance
## [1] 327.7324
##
## $AIC
## [1] 249.9147
##
## $loglikelihood
## [1] 0
##
## $null_degrees_of_freedom
## [1] 368
##
## $residual_degrees_of_freedom
## [1] 364h2o.auc(performance_h2o)## [1] 0.833657h2o.confusionMatrix(performance_h2o)## Confusion Matrix (vertical: actual; across: predicted) for max f1 @ threshold = 0.338453262465108:
## Left No Error Rate
## Left 22 38 0.633333 =38/60
## No 4 305 0.012945 =4/309
## Totals 26 343 0.113821 =42/369h2o.metric(performance_h2o)## Metrics for Thresholds: Binomial metrics as a function of classification thresholds
## threshold f1 f2 f0point5 accuracy precision recall specificity
## 1 0.999138 0.006452 0.004042 0.015974 0.165312 1.000000 0.003236 1.000000
## 2 0.998393 0.012862 0.008078 0.031546 0.168022 1.000000 0.006472 1.000000
## 3 0.998301 0.019231 0.012107 0.046729 0.170732 1.000000 0.009709 1.000000
## 4 0.998284 0.025559 0.016129 0.061538 0.173442 1.000000 0.012945 1.000000
## 5 0.998197 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 0.186603 0.918276 0.965625 0.875354 0.850949 0.848901 1.000000
## 365 0.184337 0.916914 0.965022 0.873375 0.848238 0.846575 1.000000
## 366 0.093343 0.915556 0.964419 0.871404 0.845528 0.844262 1.000000
## 367 0.080355 0.914201 0.963818 0.869443 0.842818 0.841962 1.000000
## 368 0.063383 0.912851 0.963217 0.867490 0.840108 0.839674 1.000000
## 369 0.040375 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