Goal is to automate building and tuning a classification model to predict attrition, using the h2o::h2o.automl.
Set Up Import Data *Import the cleaned data from Module 7
library(h2o)## Warning: package 'h2o' was built under R version 4.3.3##
## ----------------------------------------------------------------------
##
## 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 1.1.3 ✔ readr 2.1.4
## ✔ forcats 1.0.0 ✔ stringr 1.5.0
## ✔ ggplot2 3.4.3 ✔ tibble 3.2.1
## ✔ lubridate 1.9.2 ✔ tidyr 1.3.0
## ✔ 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 1.1.1 ──
## ✔ broom 1.0.5 ✔ rsample 1.2.0
## ✔ dials 1.2.0 ✔ tune 1.1.2
## ✔ infer 1.0.5 ✔ workflows 1.1.3
## ✔ modeldata 1.2.0 ✔ workflowsets 1.0.1
## ✔ parsnip 1.1.1 ✔ yardstick 1.2.0
## ✔ recipes 1.0.8
## ── 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()
## • Use suppressPackageStartupMessages() to eliminate package startup messageslibrary(tidyquant)## Loading required package: PerformanceAnalytics
## Loading required package: xts
## Loading required package: zoo
##
## Attaching package: 'zoo'
##
## The following objects are masked from 'package:base':
##
## as.Date, as.Date.numeric
##
##
## ######################### Warning from 'xts' package ##########################
## # #
## # The dplyr lag() function breaks how base R's lag() function is supposed to #
## # work, which breaks lag(my_xts). Calls to lag(my_xts) that you type or #
## # source() into this session won't work correctly. #
## # #
## # Use stats::lag() to make sure you're not using dplyr::lag(), or you can add #
## # conflictRules('dplyr', exclude = 'lag') to your .Rprofile to stop #
## # dplyr from breaking base R's lag() function. #
## # #
## # Code in packages is not affected. It's protected by R's namespace mechanism #
## # Set `options(xts.warn_dplyr_breaks_lag = FALSE)` to suppress this warning. #
## # #
## ###############################################################################
##
## Attaching package: 'xts'
##
## The following objects are masked from 'package:dplyr':
##
## first, last
##
##
## Attaching package: 'PerformanceAnalytics'
##
## The following object is masked from 'package:graphics':
##
## legend
##
## Loading required package: quantmod
## Loading required package: TTR
##
## Attaching package: 'TTR'
##
## The following object is masked from 'package:dials':
##
## momentum
##
## Registered S3 method overwritten by 'quantmod':
## method from
## as.zoo.data.frame zoodata <- 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.Split Data
set.seed(1234)
data_split <- initial_split(data, strata = "Attrition")
train_tbl <- training(data_split)
test_tbl <- testing(data_split)Recipes
recipe_obj <- recipe(Attrition ~ ., data = train_tbl) %>%
# Remove zero variance variables
step_zv(all_predictors())Model
# Initialize H2o
h2o.init()## Connection successful!
##
## R is connected to the H2O cluster:
## H2O cluster uptime: 1 hours 44 minutes
## H2O cluster timezone: America/New_York
## H2O data parsing timezone: UTC
## H2O cluster version: 3.44.0.3
## H2O cluster version age: 4 months and 18 days
## H2O cluster name: H2O_started_from_R_OPend_eji420
## H2O cluster total nodes: 1
## H2O cluster total memory: 3.28 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 4.3.1 (2023-06-16 ucrt)## Warning in h2o.clusterInfo():
## Your H2O cluster version is (4 months and 18 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 = 2567)##
|
| | 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)
auto_ml_models_h2o <- h2o.automl(
x = x,
y = y,
training_frame = train_h2o,
validation_frame = valid_h2o,
leaderboard_frame = test_h2o,
max_runtime_secs = 30,
exclude_algos = "DeepLearning",
# nfolds = 5,
seed = 2345
)##
|
| | 0%
|
|==== | 6%
## 13:41:34.263: 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.
## 13:41:34.263: AutoML: XGBoost is not available; skipping it.
|
|========= | 13%
|
|================ | 23%
|
|======================= | 33%
|
|========================= | 35%
|
|=========================== | 38%
|
|================================ | 45%
|
|==================================== | 52%
|
|========================================== | 59%
|
|============================================== | 66%
|
|=================================================== | 73%
|
|======================================================== | 80%
|
|======================================================================| 100%Examine the output of h2o.automl
auto_ml_models_h2o %>% typeof()## [1] "S4"auto_ml_models_h2o %>% slotNames()## [1] "project_name" "leader" "leaderboard" "event_log"
## [5] "modeling_steps" "training_info"auto_ml_models_h2o@leaderboard## model_id auc logloss
## 1 StackedEnsemble_BestOfFamily_4_AutoML_4_20240509_134134 0.8333333 0.3238283
## 2 StackedEnsemble_BestOfFamily_6_AutoML_4_20240509_134134 0.8319849 0.3242757
## 3 StackedEnsemble_BestOfFamily_3_AutoML_4_20240509_134134 0.8315534 0.3252497
## 4 StackedEnsemble_BestOfFamily_2_AutoML_4_20240509_134134 0.8314995 0.3255053
## 5 StackedEnsemble_AllModels_1_AutoML_4_20240509_134134 0.8307443 0.3256959
## 6 StackedEnsemble_AllModels_2_AutoML_4_20240509_134134 0.8300431 0.3259411
## aucpr mean_per_class_error rmse mse
## 1 0.9538473 0.3064725 0.3066953 0.09406199
## 2 0.9528834 0.3064725 0.3068768 0.09417339
## 3 0.9531354 0.3064725 0.3076645 0.09465747
## 4 0.9529747 0.3064725 0.3074807 0.09454440
## 5 0.9523748 0.3097087 0.3086613 0.09527180
## 6 0.9520009 0.3097087 0.3089480 0.09544890
##
## [44 rows x 7 columns]auto_ml_models_h2o@leader## Model Details:
## ==============
##
## H2OBinomialModel: stackedensemble
## Model ID: StackedEnsemble_BestOfFamily_4_AutoML_4_20240509_134134
## Model Summary for Stacked Ensemble:
## key value
## 1 Stacking strategy cross_validation
## 2 Number of base models (used / total) 4/4
## 3 # GBM base models (used / total) 1/1
## 4 # GLM base models (used / total) 1/1
## 5 # DRF base models (used / total) 2/2
## 6 Metalearner algorithm GLM
## 7 Metalearner fold assignment scheme Random
## 8 Metalearner nfolds 5
## 9 Metalearner fold_column NA
## 10 Custom metalearner hyperparameters None
##
##
## H2OBinomialMetrics: stackedensemble
## ** Reported on training data. **
##
## MSE: 0.04489875
## RMSE: 0.2118933
## LogLoss: 0.1672862
## Mean Per-Class Error: 0.08248189
## AUC: 0.9789573
## AUCPR: 0.9951373
## Gini: 0.9579146
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 132 22 0.142857 =22/154
## No 17 752 0.022107 =17/769
## Totals 149 774 0.042254 =39/923
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.668934 0.974725 265
## 2 max f2 0.456431 0.987417 307
## 3 max f0point5 0.737840 0.974182 245
## 4 max accuracy 0.668934 0.957746 265
## 5 max precision 0.999976 1.000000 0
## 6 max recall 0.456431 1.000000 307
## 7 max specificity 0.999976 1.000000 0
## 8 max absolute_mcc 0.668934 0.846185 265
## 9 max min_per_class_accuracy 0.777629 0.922078 226
## 10 max mean_per_class_accuracy 0.788963 0.929167 219
## 11 max tns 0.999976 154.000000 0
## 12 max fns 0.999976 617.000000 0
## 13 max fps 0.023142 154.000000 399
## 14 max tps 0.456431 769.000000 307
## 15 max tnr 0.999976 1.000000 0
## 16 max fnr 0.999976 0.802341 0
## 17 max fpr 0.023142 1.000000 399
## 18 max tpr 0.456431 1.000000 307
##
## 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: 0.06996255
## RMSE: 0.2645043
## LogLoss: 0.2984903
## Mean Per-Class Error: 0.1497896
## AUC: 0.8670407
## AUCPR: 0.9565844
## Gini: 0.7340813
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 17 6 0.260870 =6/23
## No 6 149 0.038710 =6/155
## Totals 23 155 0.067416 =12/178
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.586154 0.961290 154
## 2 max f2 0.404018 0.978535 171
## 3 max f0point5 0.586154 0.961290 154
## 4 max accuracy 0.586154 0.932584 154
## 5 max precision 0.999997 1.000000 0
## 6 max recall 0.404018 1.000000 171
## 7 max specificity 0.999997 1.000000 0
## 8 max absolute_mcc 0.586154 0.700421 154
## 9 max min_per_class_accuracy 0.784170 0.819355 130
## 10 max mean_per_class_accuracy 0.679918 0.859046 149
## 11 max tns 0.999997 23.000000 0
## 12 max fns 0.999997 154.000000 0
## 13 max fps 0.026599 23.000000 177
## 14 max tps 0.404018 155.000000 171
## 15 max tnr 0.999997 1.000000 0
## 16 max fnr 0.999997 0.993548 0
## 17 max fpr 0.026599 1.000000 177
## 18 max tpr 0.404018 1.000000 171
##
## 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: 0.09674361
## RMSE: 0.3110364
## LogLoss: 0.3288854
## Mean Per-Class Error: 0.3214455
## AUC: 0.8391696
## AUCPR: 0.9528999
## Gini: 0.6783392
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 59 95 0.616883 =95/154
## No 20 749 0.026008 =20/769
## Totals 79 844 0.124594 =115/923
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.502505 0.928704 329
## 2 max f2 0.323132 0.964979 373
## 3 max f0point5 0.709835 0.924132 249
## 4 max accuracy 0.557079 0.875406 317
## 5 max precision 0.999976 1.000000 0
## 6 max recall 0.154464 1.000000 395
## 7 max specificity 0.999976 1.000000 0
## 8 max absolute_mcc 0.709835 0.528580 249
## 9 max min_per_class_accuracy 0.833415 0.753247 175
## 10 max mean_per_class_accuracy 0.709835 0.784346 249
## 11 max tns 0.999976 154.000000 0
## 12 max fns 0.999976 735.000000 0
## 13 max fps 0.046111 154.000000 399
## 14 max tps 0.154464 769.000000 395
## 15 max tnr 0.999976 1.000000 0
## 16 max fnr 0.999976 0.955787 0
## 17 max fpr 0.046111 1.000000 399
## 18 max tpr 0.154464 1.000000 395
##
## 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 0.893984 0.005790 0.894444 0.885572 0.891892 0.901042
## auc 0.843693 0.042800 0.907728 0.791124 0.825572 0.854527
## err 0.106016 0.005790 0.105556 0.114428 0.108108 0.098958
## err_count 19.600000 2.190890 19.000000 23.000000 20.000000 19.000000
## f0point5 0.918135 0.012451 0.937500 0.903156 0.919255 0.916856
## cv_5_valid
## accuracy 0.896970
## auc 0.839513
## err 0.103030
## err_count 17.000000
## f0point5 0.913907
##
## ---
## mean sd cv_1_valid cv_2_valid cv_3_valid
## precision 0.905055 0.020884 0.938776 0.882979 0.907975
## r2 0.306485 0.051873 0.368996 0.232403 0.293534
## recall 0.976079 0.026917 0.932432 0.994012 0.967320
## residual_deviance 120.399220 16.393864 106.648430 145.820110 127.008804
## rmse 0.310082 0.013127 0.303703 0.328449 0.317903
## specificity 0.477511 0.151029 0.718750 0.352941 0.531250
## cv_4_valid cv_5_valid
## precision 0.899441 0.896104
## r2 0.339947 0.297544
## recall 0.993827 0.992806
## residual_deviance 115.065250 107.453514
## rmse 0.294989 0.305365
## specificity 0.400000 0.384615best_model <- auto_ml_models_h2o@leaderSave and Load
?h2o.getModel## starting httpd help server ... done?h2o.saveModel
?h2o.loadModel
# h2o.getModel("GLM_1_AutoML_4_20240423_111307") %>%
# h2o.saveModel("h2o_models/")
# best_model <- h2o.loadModel("h2o_models/GLM_1_AutoML_4_20240423_111307")Make Predictions
predictions <- 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 Left 0.584 0.416 41 Left Travel_Rarely 1102 Sales
## 2 No 0.0214 0.979 49 No Travel_Frequently 279 Research & …
## 3 No 0.253 0.747 33 No Travel_Frequently 1392 Research & …
## 4 No 0.144 0.856 59 No Travel_Rarely 1324 Research & …
## 5 No 0.0733 0.927 38 No Travel_Frequently 216 Research & …
## 6 No 0.291 0.709 29 No Travel_Rarely 153 Research & …
## 7 No 0.0367 0.963 34 No Travel_Rarely 1346 Research & …
## 8 Left 0.873 0.127 28 Left Travel_Rarely 103 Research & …
## 9 No 0.275 0.725 22 No Non-Travel 1123 Research & …
## 10 No 0.0236 0.976 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>, …Evaluate Model
?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_4_20240509_134134"
##
## $model$type
## [1] "Key<Model>"
##
## $model$URL
## [1] "/3/Models/StackedEnsemble_BestOfFamily_4_AutoML_4_20240509_134134"
##
##
## $model_checksum
## [1] "2117851016217035208"
##
## $frame
## $frame$name
## [1] "test_tbl_sid_8c13_3"
##
##
## $frame_checksum
## [1] "-54192601206779456"
##
## $description
## NULL
##
## $scoring_time
## [1] 1.715277e+12
##
## $predictions
## NULL
##
## $MSE
## [1] 0.09406199
##
## $RMSE
## [1] 0.3066953
##
## $nobs
## [1] 369
##
## $custom_metric_name
## NULL
##
## $custom_metric_value
## [1] 0
##
## $r2
## [1] 0.3091923
##
## $logloss
## [1] 0.3238283
##
## $AUC
## [1] 0.8333333
##
## $pr_auc
## [1] 0.9538473
##
## $Gini
## [1] 0.6666667
##
## $mean_per_class_error
## [1] 0.3064725
##
## $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 24 36 0.6000 = 36 / 60
## No 4 305 0.0129 = 4 / 309
## Totals 28 341 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 0.999998 0.006452 0.004042 0.015974 0.165312 1.000000 0.003236 1.000000
## 2 0.999998 0.012862 0.008078 0.031546 0.168022 1.000000 0.006472 1.000000
## 3 0.999997 0.019231 0.012107 0.046729 0.170732 1.000000 0.009709 1.000000
## 4 0.999996 0.025559 0.016129 0.061538 0.173442 1.000000 0.012945 1.000000
## 5 0.999996 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 0.234042 0.918276 0.965625 0.875354 0.850949 0.848901 1.000000
## 365 0.158869 0.916914 0.965022 0.873375 0.848238 0.846575 1.000000
## 366 0.126905 0.915556 0.964419 0.871404 0.845528 0.844262 1.000000
## 367 0.124947 0.914201 0.963818 0.869443 0.842818 0.841962 1.000000
## 368 0.048284 0.912851 0.963217 0.867490 0.840108 0.839674 1.000000
## 369 0.045413 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 0.435557 0.938462 340
## 2 max f2 0.340441 0.970477 355
## 3 max f0point5 0.666511 0.922330 308
## 4 max accuracy 0.528663 0.891599 328
## 5 max precision 0.999998 1.000000 0
## 6 max recall 0.340441 1.000000 355
## 7 max specificity 0.999998 1.000000 0
## 8 max absolute_mcc 0.528663 0.555055 328
## 9 max min_per_class_accuracy 0.840259 0.750000 246
## 10 max mean_per_class_accuracy 0.675604 0.764644 304
## 11 max tns 0.999998 60.000000 0
## 12 max fns 0.999998 308.000000 0
## 13 max fps 0.045413 60.000000 368
## 14 max tps 0.340441 309.000000 355
## 15 max tnr 0.999998 1.000000 0
## 16 max fnr 0.999998 0.996764 0
## 17 max fpr 0.045413 1.000000 368
## 18 max tpr 0.340441 1.000000 355
##
## $gains_lift_table
## Gains/Lift Table: Avg response rate: 83.74 %, avg score: 83.08 %
## group cumulative_data_fraction lower_threshold lift cumulative_lift
## 1 1 0.01084011 0.999996 1.194175 1.194175
## 2 2 0.02168022 0.999994 1.194175 1.194175
## 3 3 0.03252033 0.999989 1.194175 1.194175
## 4 4 0.04065041 0.999983 1.194175 1.194175
## 5 5 0.05149051 0.999971 1.194175 1.194175
## 6 6 0.10027100 0.991533 1.194175 1.194175
## 7 7 0.15176152 0.987561 1.068472 1.151526
## 8 8 0.20054201 0.979202 1.127832 1.145762
## 9 9 0.30081301 0.966820 1.161900 1.151141
## 10 10 0.40108401 0.941108 1.129625 1.145762
## 11 11 0.50135501 0.911912 1.097350 1.136080
## 12 12 0.59891599 0.872145 1.127832 1.134736
## 13 13 0.69918699 0.807201 0.968250 1.110860
## 14 14 0.79945799 0.709356 1.032800 1.101070
## 15 15 0.89972900 0.507787 0.871425 1.075477
## 16 16 1.00000000 0.045413 0.322750 1.000000
## response_rate score cumulative_response_rate cumulative_score
## 1 1.000000 0.999997 1.000000 0.999997
## 2 1.000000 0.999996 1.000000 0.999997
## 3 1.000000 0.999991 1.000000 0.999995
## 4 1.000000 0.999988 1.000000 0.999993
## 5 1.000000 0.999979 1.000000 0.999990
## 6 1.000000 0.995097 1.000000 0.997610
## 7 0.894737 0.989513 0.964286 0.994863
## 8 0.944444 0.983520 0.959459 0.992104
## 9 0.972973 0.972680 0.963964 0.985629
## 10 0.945946 0.953924 0.959459 0.977703
## 11 0.918919 0.928769 0.951351 0.967916
## 12 0.944444 0.892965 0.950226 0.955707
## 13 0.810811 0.845607 0.930233 0.939917
## 14 0.864865 0.763413 0.922034 0.917779
## 15 0.729730 0.624195 0.900602 0.885061
## 16 0.270270 0.344003 0.837398 0.830808
## 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 0.058252 0.119741 19.417476 19.417476
## 7 0.055016 0.174757 6.847215 15.152566
## 8 0.055016 0.229773 12.783172 14.576227
## 9 0.116505 0.346278 16.189976 15.114143
## 10 0.113269 0.459547 12.962477 14.576227
## 11 0.110032 0.569579 9.734978 13.607977
## 12 0.110032 0.679612 12.783172 13.473619
## 13 0.097087 0.776699 -3.175020 11.086024
## 14 0.103560 0.880259 3.279979 10.106961
## 15 0.087379 0.967638 -12.857518 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 0.119741
## 7 0.141424
## 8 0.179773
## 9 0.279612
## 10 0.359547
## 11 0.419579
## 12 0.496278
## 13 0.476699
## 14 0.496926
## 15 0.417638
## 16 0.000000
##
## $residual_deviance
## [1] 238.9853
##
## $null_deviance
## [1] 327.6898
##
## $AIC
## [1] 248.9853
##
## $loglikelihood
## [1] 0
##
## $null_degrees_of_freedom
## [1] 368
##
## $residual_degrees_of_freedom
## [1] 364h2o.auc(performance_h2o)## [1] 0.8333333h2o.confusionMatrix(performance_h2o)## Confusion Matrix (vertical: actual; across: predicted) for max f1 @ threshold = 0.435557215433683:
## Left No Error Rate
## Left 24 36 0.600000 =36/60
## No 4 305 0.012945 =4/309
## Totals 28 341 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 0.999998 0.006452 0.004042 0.015974 0.165312 1.000000 0.003236 1.000000
## 2 0.999998 0.012862 0.008078 0.031546 0.168022 1.000000 0.006472 1.000000
## 3 0.999997 0.019231 0.012107 0.046729 0.170732 1.000000 0.009709 1.000000
## 4 0.999996 0.025559 0.016129 0.061538 0.173442 1.000000 0.012945 1.000000
## 5 0.999996 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 0.234042 0.918276 0.965625 0.875354 0.850949 0.848901 1.000000
## 365 0.158869 0.916914 0.965022 0.873375 0.848238 0.846575 1.000000
## 366 0.126905 0.915556 0.964419 0.871404 0.845528 0.844262 1.000000
## 367 0.124947 0.914201 0.963818 0.869443 0.842818 0.841962 1.000000
## 368 0.048284 0.912851 0.963217 0.867490 0.840108 0.839674 1.000000
## 369 0.045413 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