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 1.1.3 ✔ readr 2.1.4
## ✔ forcats 1.0.0 ✔ stringr 1.5.0
## ✔ ggplot2 3.4.3 ✔ tibble 3.2.1
## ✔ lubridate 1.9.3 ✔ 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 1.0.6 ✔ workflows 1.1.3
## ✔ modeldata 1.3.0 ✔ workflowsets 1.0.1
## ✔ parsnip 1.1.1 ✔ yardstick 1.3.0
## ✔ 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()
## • Search for functions across packages at https://www.tidymodels.org/find/library(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.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: 7 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: 4 months and 9 days
## H2O cluster name: H2O_started_from_R_Vanessa_vmr042
## H2O cluster total nodes: 1
## H2O cluster total memory: 0.99 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 4.3.1 (2023-06-16)## Warning in h2o.clusterInfo():
## Your H2O cluster version is (4 months and 9 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.html# Split training
split.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)
models_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 = 2345
)##
|
| | 0%
|
|=== | 4%
## 11:38:38.567: 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.
|
|======== | 12%
|
|============== | 20%
|
|=================== | 27%
|
|======================== | 34%
|
|============================ | 41%
|
|================================= | 48%
|
|====================================== | 55%
|
|=========================================== | 62%
|
|================================================ | 69%
|
|===================================================== | 76%
|
|=========================================================== | 84%
|
|===================================================================== | 99%
|
|======================================================================| 100%Examine 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_19_20240430_113838 0.8295038 0.3281184
## 2 GLM_1_AutoML_19_20240430_113838 0.8269687 0.3310361
## 3 StackedEnsemble_BestOfFamily_3_AutoML_19_20240430_113838 0.8242179 0.3323141
## 4 StackedEnsemble_BestOfFamily_2_AutoML_19_20240430_113838 0.8236246 0.3333811
## 5 StackedEnsemble_AllModels_2_AutoML_19_20240430_113838 0.8209817 0.3325003
## 6 XGBoost_1_AutoML_19_20240430_113838 0.8194714 0.3426538
## aucpr mean_per_class_error rmse mse
## 1 0.9515786 0.2946602 0.3097197 0.09592630
## 2 0.9469916 0.2796117 0.3081993 0.09498682
## 3 0.9478006 0.3013754 0.3106001 0.09647245
## 4 0.9458118 0.3013754 0.3102831 0.09627562
## 5 0.9475893 0.3097087 0.3114966 0.09703014
## 6 0.9479057 0.2877023 0.3186393 0.10153097
##
## [29 rows x 7 columns]models_h2o@leader## Model Details:
## ==============
##
## H2OBinomialModel: stackedensemble
## Model ID: StackedEnsemble_BestOfFamily_1_AutoML_19_20240430_113838
## Model Summary for Stacked Ensemble:
## key value
## 1 Stacking strategy cross_validation
## 2 Number of base models (used / total) 3/3
## 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 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: 0.07243405
## RMSE: 0.2691357
## LogLoss: 0.2560715
## Mean Per-Class Error: 0.2006612
## AUC: 0.9131483
## AUCPR: 0.9752321
## Gini: 0.8262966
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 95 59 0.383117 =59/154
## No 14 755 0.018205 =14/769
## Totals 109 814 0.079090 =73/923
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.563305 0.953885 300
## 2 max f2 0.387781 0.973316 339
## 3 max f0point5 0.634497 0.945223 279
## 4 max accuracy 0.612028 0.920910 286
## 5 max precision 0.998821 1.000000 0
## 6 max recall 0.307987 1.000000 356
## 7 max specificity 0.998821 1.000000 0
## 8 max absolute_mcc 0.612028 0.698562 286
## 9 max min_per_class_accuracy 0.805352 0.831169 196
## 10 max mean_per_class_accuracy 0.749296 0.846035 230
## 11 max tns 0.998821 154.000000 0
## 12 max fns 0.998821 767.000000 0
## 13 max fps 0.043550 154.000000 399
## 14 max tps 0.307987 769.000000 356
## 15 max tnr 0.998821 1.000000 0
## 16 max fnr 0.998821 0.997399 0
## 17 max fpr 0.043550 1.000000 399
## 18 max tpr 0.307987 1.000000 356
##
## 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: 0.06682641
## RMSE: 0.258508
## LogLoss: 0.2484445
## Mean Per-Class Error: 0.1835905
## AUC: 0.8824684
## AUCPR: 0.9741389
## Gini: 0.7649369
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 15 8 0.347826 =8/23
## No 3 152 0.019355 =3/155
## Totals 18 160 0.061798 =11/178
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.555140 0.965079 159
## 2 max f2 0.481337 0.983503 167
## 3 max f0point5 0.650284 0.963303 151
## 4 max accuracy 0.555140 0.938202 159
## 5 max precision 0.997148 1.000000 0
## 6 max recall 0.481337 1.000000 167
## 7 max specificity 0.997148 1.000000 0
## 8 max absolute_mcc 0.555140 0.704066 159
## 9 max min_per_class_accuracy 0.810037 0.819355 130
## 10 max mean_per_class_accuracy 0.650284 0.865498 151
## 11 max tns 0.997148 23.000000 0
## 12 max fns 0.997148 154.000000 0
## 13 max fps 0.031089 23.000000 177
## 14 max tps 0.481337 155.000000 167
## 15 max tnr 0.997148 1.000000 0
## 16 max fnr 0.997148 0.993548 0
## 17 max fpr 0.031089 1.000000 177
## 18 max tpr 0.481337 1.000000 167
##
## 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: 0.09599907
## RMSE: 0.3098372
## LogLoss: 0.3301214
## Mean Per-Class Error: 0.3045615
## AUC: 0.8371304
## AUCPR: 0.9469827
## Gini: 0.6742607
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Left No Error Rate
## Left 64 90 0.584416 =90/154
## No 19 750 0.024707 =19/769
## Totals 83 840 0.118093 =109/923
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.514072 0.932256 328
## 2 max f2 0.317557 0.965995 373
## 3 max f0point5 0.744478 0.919350 236
## 4 max accuracy 0.514072 0.881907 328
## 5 max precision 0.998861 1.000000 0
## 6 max recall 0.220023 1.000000 387
## 7 max specificity 0.998861 1.000000 0
## 8 max absolute_mcc 0.661111 0.517520 276
## 9 max min_per_class_accuracy 0.833929 0.754226 182
## 10 max mean_per_class_accuracy 0.744478 0.778486 236
## 11 max tns 0.998861 154.000000 0
## 12 max fns 0.998861 766.000000 0
## 13 max fps 0.059958 154.000000 399
## 14 max tps 0.220023 769.000000 387
## 15 max tnr 0.998861 1.000000 0
## 16 max fnr 0.998861 0.996099 0
## 17 max fpr 0.059958 1.000000 399
## 18 max tpr 0.220023 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 0.886466 0.014695 0.876405 0.890000 0.867021 0.895288
## auc 0.838469 0.022919 0.862905 0.806471 0.858289 0.829881
## err 0.113534 0.014695 0.123596 0.110000 0.132979 0.104712
## err_count 21.000000 3.316625 22.000000 22.000000 25.000000 20.000000
## f0point5 0.908853 0.012312 0.899873 0.906183 0.900243 0.908046
## cv_5_valid
## accuracy 0.903614
## auc 0.834800
## err 0.096386
## err_count 16.000000
## f0point5 0.929919
##
## ---
## mean sd cv_1_valid cv_2_valid cv_3_valid
## precision 0.892254 0.015649 0.881988 0.885417 0.886228
## r2 0.306176 0.042649 0.340557 0.251995 0.333107
## recall 0.982436 0.017299 0.979310 1.000000 0.961039
## residual_deviance 121.534590 13.034665 117.335590 133.204570 125.354195
## rmse 0.309476 0.005866 0.315580 0.308821 0.314318
## specificity 0.405205 0.086548 0.424242 0.266667 0.441176
## cv_4_valid cv_5_valid
## precision 0.887641 0.920000
## r2 0.337306 0.267915
## recall 1.000000 0.971831
## residual_deviance 130.846920 100.931660
## rmse 0.307758 0.300900
## specificity 0.393939 0.500000best_model <- models_h2o@leader?h2o.getModel
?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")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)## # A tibble: 369 × 35
## predict Left No Age Attrition BusinessTravel DailyRate Department
## <fct> <dbl> <dbl> <dbl> <fct> <fct> <dbl> <fct>
## 1 No 0.436 0.564 41 Left Travel_Rarely 1102 Sales
## 2 No 0.0234 0.977 49 No Travel_Frequently 279 Research & …
## 3 No 0.391 0.609 33 No Travel_Frequently 1392 Research & …
## 4 No 0.251 0.749 59 No Travel_Rarely 1324 Research & …
## 5 No 0.0807 0.919 38 No Travel_Frequently 216 Research & …
## 6 No 0.368 0.632 29 No Travel_Rarely 153 Research & …
## 7 No 0.0431 0.957 34 No Travel_Rarely 1346 Research & …
## 8 Left 0.888 0.112 28 Left Travel_Rarely 103 Research & …
## 9 No 0.310 0.690 22 No Non-Travel 1123 Research & …
## 10 No 0.0308 0.969 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_19_20240430_113838"
##
## $model$type
## [1] "Key<Model>"
##
## $model$URL
## [1] "/3/Models/StackedEnsemble_BestOfFamily_1_AutoML_19_20240430_113838"
##
##
## $model_checksum
## [1] "-4116832895039817680"
##
## $frame
## $frame$name
## [1] "test_tbl_sid_bf5c_3"
##
##
## $frame_checksum
## [1] "-54192601206779456"
##
## $description
## NULL
##
## $scoring_time
## [1] 1.714492e+12
##
## $predictions
## NULL
##
## $MSE
## [1] 0.0959263
##
## $RMSE
## [1] 0.3097197
##
## $nobs
## [1] 369
##
## $custom_metric_name
## NULL
##
## $custom_metric_value
## [1] 0
##
## $r2
## [1] 0.2955005
##
## $logloss
## [1] 0.3281184
##
## $AUC
## [1] 0.8295038
##
## $pr_auc
## [1] 0.9515786
##
## $Gini
## [1] 0.6590076
##
## $mean_per_class_error
## [1] 0.2946602
##
## $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 26 34 0.5667 = 34 / 60
## No 7 302 0.0227 = 7 / 309
## Totals 33 336 0.1111 = 41 / 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 0.998336 0.006452 0.004042 0.015974 0.165312 1.000000 0.003236 1.000000
## 2 0.998329 0.012862 0.008078 0.031546 0.168022 1.000000 0.006472 1.000000
## 3 0.998118 0.019231 0.012107 0.046729 0.170732 1.000000 0.009709 1.000000
## 4 0.997497 0.025559 0.016129 0.061538 0.173442 1.000000 0.012945 1.000000
## 5 0.997362 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 0.193350 0.918276 0.965625 0.875354 0.850949 0.848901 1.000000
## 365 0.162010 0.916914 0.965022 0.873375 0.848238 0.846575 1.000000
## 366 0.131237 0.915556 0.964419 0.871404 0.845528 0.844262 1.000000
## 367 0.112196 0.914201 0.963818 0.869443 0.842818 0.841962 1.000000
## 368 0.042625 0.912851 0.963217 0.867490 0.840108 0.839674 1.000000
## 369 0.040768 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 0.431163 0.936434 335
## 2 max f2 0.333800 0.969163 352
## 3 max f0point5 0.608785 0.918627 315
## 4 max accuracy 0.431163 0.888889 335
## 5 max precision 0.998336 1.000000 0
## 6 max recall 0.227134 1.000000 360
## 7 max specificity 0.998336 1.000000 0
## 8 max absolute_mcc 0.431163 0.531045 335
## 9 max min_per_class_accuracy 0.840526 0.750000 246
## 10 max mean_per_class_accuracy 0.814673 0.759628 264
## 11 max tns 0.998336 60.000000 0
## 12 max fns 0.998336 308.000000 0
## 13 max fps 0.040768 60.000000 368
## 14 max tps 0.227134 309.000000 360
## 15 max tnr 0.998336 1.000000 0
## 16 max fnr 0.998336 0.996764 0
## 17 max fpr 0.040768 1.000000 368
## 18 max tpr 0.227134 1.000000 360
##
## $gains_lift_table
## Gains/Lift Table: Avg response rate: 83.74 %, avg score: 82.94 %
## group cumulative_data_fraction lower_threshold lift cumulative_lift
## 1 1 0.01084011 0.997405 1.194175 1.194175
## 2 2 0.02168022 0.996422 1.194175 1.194175
## 3 3 0.03252033 0.995477 1.194175 1.194175
## 4 4 0.04065041 0.993970 1.194175 1.194175
## 5 5 0.05149051 0.992949 1.194175 1.194175
## 6 6 0.10027100 0.988886 1.127832 1.161900
## 7 7 0.15176152 0.984330 1.131323 1.151526
## 8 8 0.20054201 0.979802 1.194175 1.161900
## 9 9 0.30081301 0.967011 1.129625 1.151141
## 10 10 0.40108401 0.949897 1.129625 1.145762
## 11 11 0.50135501 0.919002 1.097350 1.136080
## 12 12 0.59891599 0.880412 1.127832 1.134736
## 13 13 0.69918699 0.818078 1.000525 1.115489
## 14 14 0.79945799 0.714733 1.000525 1.101070
## 15 15 0.89972900 0.489309 0.871425 1.075477
## 16 16 1.00000000 0.040768 0.322750 1.000000
## response_rate score cumulative_response_rate cumulative_score
## 1 1.000000 0.998070 1.000000 0.998070
## 2 1.000000 0.997113 1.000000 0.997591
## 3 1.000000 0.995820 1.000000 0.997001
## 4 1.000000 0.994451 1.000000 0.996491
## 5 1.000000 0.993601 1.000000 0.995882
## 6 0.944444 0.990790 0.972973 0.993405
## 7 0.947368 0.987058 0.964286 0.991252
## 8 1.000000 0.982354 0.972973 0.989087
## 9 0.945946 0.972292 0.963964 0.983489
## 10 0.945946 0.957669 0.959459 0.977034
## 11 0.918919 0.937192 0.951351 0.969065
## 12 0.944444 0.896925 0.950226 0.957314
## 13 0.837838 0.851016 0.934109 0.942070
## 14 0.837838 0.778395 0.922034 0.921541
## 15 0.729730 0.610855 0.900602 0.886916
## 16 0.270270 0.313308 0.837398 0.829400
## 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 0.058252 0.233010 19.417476 16.189976
## 9 0.113269 0.346278 12.962477 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 0.100324 0.779935 0.052480 11.548882
## 14 0.100324 0.880259 0.052480 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 0.099838
## 7 0.141424
## 8 0.199676
## 9 0.279612
## 10 0.359547
## 11 0.419579
## 12 0.496278
## 13 0.496602
## 14 0.496926
## 15 0.417638
## 16 0.000000
##
## $residual_deviance
## [1] 242.1514
##
## $null_deviance
## [1] 327.6898
##
## $AIC
## [1] 250.1514
##
## $loglikelihood
## [1] 0
##
## $null_degrees_of_freedom
## [1] 368
##
## $residual_degrees_of_freedom
## [1] 365h2o.auc(best_model)## [1] 0.9131483h2o.confusionMatrix(performance_h2o)## Confusion Matrix (vertical: actual; across: predicted) for max f1 @ threshold = 0.431163495747449:
## Left No Error Rate
## Left 26 34 0.566667 =34/60
## No 7 302 0.022654 =7/309
## Totals 33 336 0.111111 =41/369h2o.metric(performance_h2o) %>% as_tibble() %>% filter(threshold %>% between(0.43, 0.44))## # A tibble: 1 × 20
## threshold f1 f2 f0point5 accuracy precision recall specificity
## <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 0.431 0.936 0.961 0.913 0.889 0.899 0.977 0.433
## # ℹ 12 more variables: absolute_mcc <dbl>, min_per_class_accuracy <dbl>,
## # mean_per_class_accuracy <dbl>, tns <dbl>, fns <dbl>, fps <dbl>, tps <dbl>,
## # tnr <dbl>, fnr <dbl>, fpr <dbl>, tpr <dbl>, idx <int>