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 1.0.7 ✔ rsample 1.2.1
## ✔ dials 1.2.1 ✔ 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 1.3.1
## ✔ recipes 1.1.0## Warning: package 'broom' was built under R version 4.3.3## Warning: package 'modeldata' was built under R version 4.3.3## Warning: package 'recipes' was built under R version 4.3.3## ── 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()
## • Dig deeper into tidy modeling with R at https://www.tmwr.orglibrary(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))## New names:
## Rows: 1470 Columns: 33
## ── Column specification
## ──────────────────────────────────────────────────────── Delimiter: "," chr
## (8): Attrition, BusinessTravel, Department, EducationField, Gender, Job... dbl
## (25): ...1, Age, DailyRate, DistanceFromHome, Education, EmployeeNumber,...
## 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.
## • `` -> `...1`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: 4 hours 56 minutes
## H2O cluster timezone: America/New_York
## H2O data parsing timezone: UTC
## H2O cluster version: 3.44.0.3
## H2O cluster version age: 10 months and 30 days
## H2O cluster name: H2O_started_from_R_jordanlanowy_fhp551
## H2O cluster total nodes: 1
## H2O cluster total memory: 1.30 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 4.3.2 (2023-10-31)## Warning in h2o.clusterInfo():
## Your H2O cluster version is (10 months and 30 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 = 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%
|
|== | 3%
## 17:56:20.765: 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%
|
|============== | 21%
|
|==================== | 28%
|
|======================================================================| 100%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_20241119_175620 0.8482740 0.3094056
## 2 StackedEnsemble_AllModels_1_AutoML_11_20241119_175620 0.8451996 0.3148845
## 3 GLM_1_AutoML_11_20241119_175620 0.8399137 0.3186238
## 4 GBM_1_AutoML_11_20241119_175620 0.8374326 0.3322943
## 5 XGBoost_2_AutoML_11_20241119_175620 0.8278857 0.3299205
## 6 XGBoost_1_AutoML_11_20241119_175620 0.8270766 0.3411288
## aucpr mean_per_class_error rmse mse
## 1 0.6530560 0.2133495 0.3006395 0.09038409
## 2 0.6165789 0.2339806 0.3040087 0.09242130
## 3 0.6509898 0.2557443 0.3064529 0.09391340
## 4 0.5820366 0.2418285 0.3163850 0.10009950
## 5 0.6008391 0.2441748 0.3092394 0.09562904
## 6 0.5284256 0.2654531 0.3208495 0.10294437
##
## [12 rows x 7 columns]models_h2o@leader## Model Details:
## ==============
##
## H2OBinomialModel: stackedensemble
## Model ID: StackedEnsemble_BestOfFamily_1_AutoML_11_20241119_175620
## 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: 0.0489028
## RMSE: 0.2211398
## LogLoss: 0.1843305
## Mean Per-Class Error: 0.1157107
## AUC: 0.9671277
## AUCPR: 0.91407
## Gini: 0.9342555
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## No Yes Error Rate
## No 779 9 0.011421 =9/788
## Yes 33 117 0.220000 =33/150
## Totals 812 126 0.044776 =42/938
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.388860 0.847826 109
## 2 max f2 0.227183 0.862944 158
## 3 max f0point5 0.407536 0.904762 103
## 4 max accuracy 0.407536 0.955224 103
## 5 max precision 0.965939 1.000000 0
## 6 max recall 0.049723 1.000000 314
## 7 max specificity 0.965939 1.000000 0
## 8 max absolute_mcc 0.388860 0.826102 109
## 9 max min_per_class_accuracy 0.206539 0.914975 170
## 10 max mean_per_class_accuracy 0.227183 0.920338 158
## 11 max tns 0.965939 788.000000 0
## 12 max fns 0.965939 149.000000 0
## 13 max fps 0.000130 788.000000 399
## 14 max tps 0.049723 150.000000 314
## 15 max tnr 0.965939 1.000000 0
## 16 max fnr 0.965939 0.993333 0
## 17 max fpr 0.000130 1.000000 399
## 18 max tpr 0.049723 1.000000 314
##
## 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: 0.09471161
## RMSE: 0.3077525
## LogLoss: 0.3085773
## Mean Per-Class Error: 0.1996187
## AUC: 0.8651961
## AUCPR: 0.6140732
## Gini: 0.7303922
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## No Yes Error Rate
## No 122 14 0.102941 =14/136
## Yes 8 19 0.296296 =8/27
## Totals 130 33 0.134969 =22/163
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.300237 0.633333 32
## 2 max f2 0.277669 0.689655 36
## 3 max f0point5 0.476573 0.631579 16
## 4 max accuracy 0.476573 0.877301 16
## 5 max precision 0.966238 1.000000 0
## 6 max recall 0.035832 1.000000 119
## 7 max specificity 0.966238 1.000000 0
## 8 max absolute_mcc 0.300237 0.555809 32
## 9 max min_per_class_accuracy 0.220842 0.777778 44
## 10 max mean_per_class_accuracy 0.277669 0.807870 36
## 11 max tns 0.966238 136.000000 0
## 12 max fns 0.966238 26.000000 0
## 13 max fps 0.000099 136.000000 162
## 14 max tps 0.035832 27.000000 119
## 15 max tnr 0.966238 1.000000 0
## 16 max fnr 0.966238 0.962963 0
## 17 max fpr 0.000099 1.000000 162
## 18 max tpr 0.035832 1.000000 119
##
## 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: 0.09528276
## RMSE: 0.3086791
## LogLoss: 0.3328032
## Mean Per-Class Error: 0.2453723
## AUC: 0.8242851
## AUCPR: 0.608214
## Gini: 0.6485702
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## No Yes Error Rate
## No 727 61 0.077411 =61/788
## Yes 62 88 0.413333 =62/150
## Totals 789 149 0.131130 =123/938
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.309176 0.588629 113
## 2 max f2 0.189945 0.630734 180
## 3 max f0point5 0.482884 0.628019 53
## 4 max accuracy 0.482884 0.880597 53
## 5 max precision 0.972574 1.000000 0
## 6 max recall 0.000738 1.000000 398
## 7 max specificity 0.972574 1.000000 0
## 8 max absolute_mcc 0.309176 0.510638 113
## 9 max min_per_class_accuracy 0.160674 0.746667 199
## 10 max mean_per_class_accuracy 0.215192 0.764340 164
## 11 max tns 0.972574 788.000000 0
## 12 max fns 0.972574 149.000000 0
## 13 max fps 0.000094 788.000000 399
## 14 max tps 0.000738 150.000000 398
## 15 max tnr 0.972574 1.000000 0
## 16 max fnr 0.972574 0.993333 0
## 17 max fpr 0.000094 1.000000 399
## 18 max tpr 0.000738 1.000000 398
##
## 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 0.844706 0.066918 0.904306 0.880829 0.850000 0.857143
## auc 0.821823 0.039724 0.853186 0.764524 0.864662 0.819169
## err 0.155294 0.066918 0.095694 0.119171 0.150000 0.142857
## err_count 28.200000 8.871302 20.000000 23.000000 27.000000 28.000000
## f0point5 0.564303 0.115758 0.741758 0.547445 0.543478 0.570652
## cv_5_valid
## accuracy 0.731250
## auc 0.807574
## err 0.268750
## err_count 43.000000
## f0point5 0.418182
##
## ---
## mean sd cv_1_valid cv_2_valid cv_3_valid
## precision 0.544427 0.135546 0.750000 0.535714 0.512821
## r2 0.276706 0.077089 0.369137 0.212622 0.336482
## recall 0.706583 0.093637 0.710526 0.600000 0.714286
## residual_deviance 124.462150 13.618898 141.060060 125.348150 106.857640
## rmse 0.310251 0.016827 0.306345 0.297961 0.295226
## specificity 0.869619 0.095023 0.947368 0.922619 0.875000
## cv_4_valid cv_5_valid
## precision 0.552632 0.370968
## r2 0.275453 0.189838
## recall 0.656250 0.851852
## residual_deviance 133.345600 115.699300
## rmse 0.314611 0.337112
## specificity 0.896341 0.706767?h2o.getModel
?h2o.saveModel
?h2o.loadModel
h2o.getModel("StackedEnsemble_BestOfFamily_1_AutoML_9_20241119_174925") %>%
h2o.saveModel("h2o_models/", force = TRUE)## [1] "/Users/jordanlanowy/Desktop/PSU_DAT3100/11_module13/h2o_models/StackedEnsemble_BestOfFamily_1_AutoML_9_20241119_174925"best_model <- h2o.loadModel("h2o_models/StackedEnsemble_BestOfFamily_1_AutoML_9_20241119_174925")predictions <- h2o.predict(best_model, newdata = test_h2o)##
|
| | 0%
|
|======================================================================| 100%predictions_tbl <- predictions %>%
as.tibble()## Warning: `as.tibble()` was deprecated in tibble 2.0.0.
## ℹ Please use `as_tibble()` instead.
## ℹ The signature and semantics have changed, see `?as_tibble`.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.predictions_tbl %>%
bind_cols(test_tbl)## New names:
## • `...1` -> `...4`## # A tibble: 369 × 36
## predict No Yes ...4 Age Attrition BusinessTravel DailyRate
## <fct> <dbl> <dbl> <dbl> <dbl> <fct> <fct> <dbl>
## 1 Yes 0.657 0.343 7 59 No Travel_Rarely 1324
## 2 No 0.893 0.107 11 35 No Travel_Rarely 809
## 3 No 0.931 0.0687 14 34 No Travel_Rarely 1346
## 4 Yes 0.565 0.435 18 22 No Non-Travel 1123
## 5 No 0.980 0.0199 19 53 No Travel_Rarely 1219
## 6 No 0.969 0.0307 21 24 No Non-Travel 673
## 7 No 0.787 0.213 24 21 No Travel_Rarely 391
## 8 No 0.891 0.109 25 34 Yes Travel_Rarely 699
## 9 No 0.994 0.00636 26 53 No Travel_Rarely 1282
## 10 Yes 0.173 0.827 27 32 Yes Travel_Frequently 1125
## # ℹ 359 more rows
## # ℹ 28 more variables: Department <fct>, 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>, …?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_9_20241119_174925"
##
## $model$type
## [1] "Key<Model>"
##
## $model$URL
## [1] "/3/Models/StackedEnsemble_BestOfFamily_1_AutoML_9_20241119_174925"
##
##
## $model_checksum
## [1] "7206720806778980384"
##
## $frame
## $frame$name
## [1] "test_tbl_sid_bc2c_3"
##
##
## $frame_checksum
## [1] 3.637474e+13
##
## $description
## NULL
##
## $scoring_time
## [1] 1.732057e+12
##
## $predictions
## NULL
##
## $MSE
## [1] 0.09038409
##
## $RMSE
## [1] 0.3006395
##
## $nobs
## [1] 369
##
## $custom_metric_name
## NULL
##
## $custom_metric_value
## [1] 0
##
## $r2
## [1] 0.3362035
##
## $logloss
## [1] 0.3094056
##
## $AUC
## [1] 0.848274
##
## $pr_auc
## [1] 0.653056
##
## $Gini
## [1] 0.696548
##
## $mean_per_class_error
## [1] 0.2133495
##
## $domain
## [1] "No" "Yes"
##
## $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
## No Yes Error Rate
## No 275 34 0.1100 = 34 / 309
## Yes 19 41 0.3167 = 19 / 60
## Totals 294 75 0.1436 = 53 / 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 0.946561 0.032787 0.020747 0.078125 0.840108 1.000000 0.016667 1.000000
## 2 0.916826 0.064516 0.041322 0.147059 0.842818 1.000000 0.033333 1.000000
## 3 0.909689 0.095238 0.061728 0.208333 0.845528 1.000000 0.050000 1.000000
## 4 0.853838 0.125000 0.081967 0.263158 0.848238 1.000000 0.066667 1.000000
## 5 0.827380 0.153846 0.102041 0.312500 0.850949 1.000000 0.083333 1.000000
## absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns fns fps tps
## 1 0.118299 0.016667 0.508333 309 59 0 1
## 2 0.167527 0.033333 0.516667 309 58 0 2
## 3 0.205458 0.050000 0.525000 309 57 0 3
## 4 0.237568 0.066667 0.533333 309 56 0 4
## 5 0.265973 0.083333 0.541667 309 55 0 5
## tnr fnr fpr tpr idx
## 1 1.000000 0.983333 0.000000 0.016667 0
## 2 1.000000 0.966667 0.000000 0.033333 1
## 3 1.000000 0.950000 0.000000 0.050000 2
## 4 1.000000 0.933333 0.000000 0.066667 3
## 5 1.000000 0.916667 0.000000 0.083333 4
##
## ---
## threshold f1 f2 f0point5 accuracy precision recall
## 364 0.003074 0.283019 0.496689 0.197889 0.176152 0.164835 1.000000
## 365 0.002808 0.282353 0.495868 0.197368 0.173442 0.164384 1.000000
## 366 0.002462 0.281690 0.495050 0.196850 0.170732 0.163934 1.000000
## 367 0.001719 0.281030 0.494234 0.196335 0.168022 0.163488 1.000000
## 368 0.001592 0.280374 0.493421 0.195822 0.165312 0.163043 1.000000
## 369 0.000644 0.279720 0.492611 0.195313 0.162602 0.162602 1.000000
## specificity absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns
## 364 0.016181 0.051645 0.016181 0.508091 5
## 365 0.012945 0.046130 0.012945 0.506472 4
## 366 0.009709 0.039895 0.009709 0.504854 3
## 367 0.006472 0.032530 0.006472 0.503236 2
## 368 0.003236 0.022971 0.003236 0.501618 1
## 369 0.000000 0.000000 0.000000 0.500000 0
## fns fps tps tnr fnr fpr tpr idx
## 364 0 304 60 0.016181 0.000000 0.983819 1.000000 363
## 365 0 305 60 0.012945 0.000000 0.987055 1.000000 364
## 366 0 306 60 0.009709 0.000000 0.990291 1.000000 365
## 367 0 307 60 0.006472 0.000000 0.993528 1.000000 366
## 368 0 308 60 0.003236 0.000000 0.996764 1.000000 367
## 369 0 309 60 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 0.250087 0.607407 74
## 2 max f2 0.220399 0.661538 84
## 3 max f0point5 0.534969 0.684524 26
## 4 max accuracy 0.534969 0.888889 26
## 5 max precision 0.946561 1.000000 0
## 6 max recall 0.011745 1.000000 334
## 7 max specificity 0.946561 1.000000 0
## 8 max absolute_mcc 0.441968 0.535812 41
## 9 max min_per_class_accuracy 0.173318 0.766667 111
## 10 max mean_per_class_accuracy 0.220399 0.790372 84
## 11 max tns 0.946561 309.000000 0
## 12 max fns 0.946561 59.000000 0
## 13 max fps 0.000644 309.000000 368
## 14 max tps 0.011745 60.000000 334
## 15 max tnr 0.946561 1.000000 0
## 16 max fnr 0.946561 0.983333 0
## 17 max fpr 0.000644 1.000000 368
## 18 max tpr 0.011745 1.000000 334
##
## $gains_lift_table
## Gains/Lift Table: Avg response rate: 16.26 %, avg score: 16.34 %
## group cumulative_data_fraction lower_threshold lift cumulative_lift
## 1 1 0.01084011 0.835847 6.150000 6.150000
## 2 2 0.02168022 0.742746 4.612500 5.381250
## 3 3 0.03252033 0.723694 6.150000 5.637500
## 4 4 0.04065041 0.646998 6.150000 5.740000
## 5 5 0.05149051 0.628951 6.150000 5.826316
## 6 6 0.10027100 0.476425 3.075000 4.487838
## 7 7 0.15176152 0.336226 2.265789 3.733929
## 8 8 0.20054201 0.250234 2.050000 3.324324
## 9 9 0.30081301 0.173418 0.831081 2.493243
## 10 10 0.40108401 0.119383 0.664865 2.036149
## 11 11 0.50135501 0.084869 0.664865 1.761892
## 12 12 0.59891599 0.057271 0.170833 1.502715
## 13 13 0.69918699 0.038274 0.498649 1.358721
## 14 14 0.79945799 0.023599 0.166216 1.209153
## 15 15 0.89972900 0.012188 0.166216 1.092922
## 16 16 1.00000000 0.000644 0.166216 1.000000
## response_rate score cumulative_response_rate cumulative_score
## 1 1.000000 0.906729 1.000000 0.906729
## 2 0.750000 0.766581 0.875000 0.836655
## 3 1.000000 0.734122 0.916667 0.802477
## 4 1.000000 0.674551 0.933333 0.776892
## 5 1.000000 0.640403 0.947368 0.748158
## 6 0.500000 0.543477 0.729730 0.648583
## 7 0.368421 0.403110 0.607143 0.565297
## 8 0.333333 0.289878 0.540541 0.498303
## 9 0.135135 0.210474 0.405405 0.402360
## 10 0.108108 0.148500 0.331081 0.338895
## 11 0.108108 0.101709 0.286486 0.291458
## 12 0.027778 0.069848 0.244344 0.255359
## 13 0.081081 0.049021 0.220930 0.225767
## 14 0.027027 0.030137 0.196610 0.201231
## 15 0.027027 0.018449 0.177711 0.180861
## 16 0.027027 0.007168 0.162602 0.163444
## capture_rate cumulative_capture_rate gain cumulative_gain
## 1 0.066667 0.066667 515.000000 515.000000
## 2 0.050000 0.116667 361.250000 438.125000
## 3 0.066667 0.183333 515.000000 463.750000
## 4 0.050000 0.233333 515.000000 474.000000
## 5 0.066667 0.300000 515.000000 482.631579
## 6 0.150000 0.450000 207.500000 348.783784
## 7 0.116667 0.566667 126.578947 273.392857
## 8 0.100000 0.666667 105.000000 232.432432
## 9 0.083333 0.750000 -16.891892 149.324324
## 10 0.066667 0.816667 -33.513514 103.614865
## 11 0.066667 0.883333 -33.513514 76.189189
## 12 0.016667 0.900000 -82.916667 50.271493
## 13 0.050000 0.950000 -50.135135 35.872093
## 14 0.016667 0.966667 -83.378378 20.915254
## 15 0.016667 0.983333 -83.378378 9.292169
## 16 0.016667 1.000000 -83.378378 0.000000
## kolmogorov_smirnov
## 1 0.066667
## 2 0.113430
## 3 0.180097
## 4 0.230097
## 5 0.296764
## 6 0.417638
## 7 0.495469
## 8 0.556634
## 9 0.536408
## 10 0.496278
## 11 0.456149
## 12 0.359547
## 13 0.299515
## 14 0.199676
## 15 0.099838
## 16 0.000000
##
## $residual_deviance
## [1] 228.3414
##
## $null_deviance
## [1] 327.6614
##
## $AIC
## [1] 238.3414
##
## $loglikelihood
## [1] 0
##
## $null_degrees_of_freedom
## [1] 368
##
## $residual_degrees_of_freedom
## [1] 364h2o.auc(performance_h2o)## [1] 0.848274h2o.confusionMatrix(performance_h2o)## Confusion Matrix (vertical: actual; across: predicted) for max f1 @ threshold = 0.250087463682527:
## No Yes Error Rate
## No 275 34 0.110032 =34/309
## Yes 19 41 0.316667 =19/60
## Totals 294 75 0.143631 =53/369h2o.metric(performance_h2o)## Metrics for Thresholds: Binomial metrics as a function of classification thresholds
## threshold f1 f2 f0point5 accuracy precision recall specificity
## 1 0.946561 0.032787 0.020747 0.078125 0.840108 1.000000 0.016667 1.000000
## 2 0.916826 0.064516 0.041322 0.147059 0.842818 1.000000 0.033333 1.000000
## 3 0.909689 0.095238 0.061728 0.208333 0.845528 1.000000 0.050000 1.000000
## 4 0.853838 0.125000 0.081967 0.263158 0.848238 1.000000 0.066667 1.000000
## 5 0.827380 0.153846 0.102041 0.312500 0.850949 1.000000 0.083333 1.000000
## absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns fns fps tps
## 1 0.118299 0.016667 0.508333 309 59 0 1
## 2 0.167527 0.033333 0.516667 309 58 0 2
## 3 0.205458 0.050000 0.525000 309 57 0 3
## 4 0.237568 0.066667 0.533333 309 56 0 4
## 5 0.265973 0.083333 0.541667 309 55 0 5
## tnr fnr fpr tpr idx
## 1 1.000000 0.983333 0.000000 0.016667 0
## 2 1.000000 0.966667 0.000000 0.033333 1
## 3 1.000000 0.950000 0.000000 0.050000 2
## 4 1.000000 0.933333 0.000000 0.066667 3
## 5 1.000000 0.916667 0.000000 0.083333 4
##
## ---
## threshold f1 f2 f0point5 accuracy precision recall
## 364 0.003074 0.283019 0.496689 0.197889 0.176152 0.164835 1.000000
## 365 0.002808 0.282353 0.495868 0.197368 0.173442 0.164384 1.000000
## 366 0.002462 0.281690 0.495050 0.196850 0.170732 0.163934 1.000000
## 367 0.001719 0.281030 0.494234 0.196335 0.168022 0.163488 1.000000
## 368 0.001592 0.280374 0.493421 0.195822 0.165312 0.163043 1.000000
## 369 0.000644 0.279720 0.492611 0.195313 0.162602 0.162602 1.000000
## specificity absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns
## 364 0.016181 0.051645 0.016181 0.508091 5
## 365 0.012945 0.046130 0.012945 0.506472 4
## 366 0.009709 0.039895 0.009709 0.504854 3
## 367 0.006472 0.032530 0.006472 0.503236 2
## 368 0.003236 0.022971 0.003236 0.501618 1
## 369 0.000000 0.000000 0.000000 0.500000 0
## fns fps tps tnr fnr fpr tpr idx
## 364 0 304 60 0.016181 0.000000 0.983819 1.000000 363
## 365 0 305 60 0.012945 0.000000 0.987055 1.000000 364
## 366 0 306 60 0.009709 0.000000 0.990291 1.000000 365
## 367 0 307 60 0.006472 0.000000 0.993528 1.000000 366
## 368 0 308 60 0.003236 0.000000 0.996764 1.000000 367
## 369 0 309 60 0.000000 0.000000 1.000000 1.000000 368