Goal is to automate building and tuning a classification model to predict climbers deaths, 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()
## • Use tidymodels_prefer() to resolve common conflicts.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 <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2020/2020-09-22/members.csv')## Rows: 76519 Columns: 21
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (10): expedition_id, member_id, peak_id, peak_name, season, sex, citizen...
## dbl (5): year, age, highpoint_metres, death_height_metres, injury_height_me...
## lgl (6): hired, success, solo, oxygen_used, died, injured
##
## 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.factors_vec <- data %>% select(year, age, highpoint_metres, death_height_metres, injury_height_metres) %>%
names()
# Treating missing values
data_clean <- data %>%
select(-death_cause, -injury_type, -death_height_metres, - injury_height_metres) %>%
drop_na() %>%
# Mutate logical Variables
mutate(died = case_when(died == "TRUE" ~ "died", died == "FALSE" ~ "no")) %>%
mutate(across(where(is.logical), as.factor)) %>%
# Recode "died"
mutate(died = if_else(died == "TRUE", "deaths", died))set.seed(1234)
data_split <- initial_split(data, strata = "died")
train_tbl <- training(data_split)
test_tbl <- testing(data_split)recipe_obj <- recipe(died ~ ., 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: 0.95 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 <- "died"
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%
|
|=== | 5%
## 11:46:12.169: 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.
## 11:46:12.735: _train param, Dropping bad and constant columns: [member_id, peak_name, death_cause, peak_id, sex, citizenship, expedition_role, season, expedition_id, injury_type]
|
|=========== | 15%
|
|================ | 23%
## 11:46:19.541: _train param, Dropping bad and constant columns: [member_id, peak_name, death_cause, peak_id, sex, citizenship, expedition_role, season, expedition_id, injury_type]
|
|===================== | 30%
|
|========================== | 38%
|
|================================ | 46%
## 11:46:26.315: _train param, Dropping bad and constant columns: [member_id, peak_name, death_cause, peak_id, sex, citizenship, expedition_role, season, expedition_id, injury_type]
|
|===================================== | 53%
|
|========================================== | 60%
|
|=============================================== | 67%
|
|==================================================== | 74%
## 11:46:35.306: _train param, Dropping unused columns: [member_id, peak_name, death_cause, peak_id, sex, citizenship, expedition_role, season, expedition_id, injury_type]
|
|========================================================= | 82%
|
|============================================================== | 89%
|
|=================================================================== | 96%
|
|======================================================================| 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
## 1 StackedEnsemble_BestOfFamily_1_AutoML_21_20240430_114612 0.9940374
## 2 GBM_1_AutoML_21_20240430_114612 0.9939811
## 3 XGBoost_1_AutoML_21_20240430_114612 0.9927011
## 4 GLM_1_AutoML_21_20240430_114612 0.6664706
## logloss aucpr mean_per_class_error rmse mse
## 1 0.011331425 0.98504766 0.01666667 0.04026143 0.0016209828
## 2 0.003175255 0.98464480 0.01666667 0.02144955 0.0004600831
## 3 0.004010885 0.98150554 0.01666667 0.02178677 0.0004746633
## 4 0.070858804 0.05070241 0.42197380 0.11725982 0.0137498658
##
## [4 rows x 7 columns]models_h2o@leader## Model Details:
## ==============
##
## H2OBinomialModel: stackedensemble
## Model ID: StackedEnsemble_BestOfFamily_1_AutoML_21_20240430_114612
## Model Summary for Stacked Ensemble:
## key value
## 1 Stacking strategy cross_validation
## 2 Number of base models (used / total) 2/3
## 3 # GBM base models (used / total) 1/1
## 4 # XGBoost base models (used / total) 1/1
## 5 # GLM base models (used / total) 0/1
## 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: 0.001482344
## RMSE: 0.03850123
## LogLoss: 0.01062543
## Mean Per-Class Error: 0.007093331
## AUC: 0.9999169
## AUCPR: 0.9958875
## Gini: 0.9998338
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## FALSE TRUE Error Rate
## FALSE 9788 1 0.000102 =1/9789
## TRUE 2 140 0.014085 =2/142
## Totals 9790 141 0.000302 =3/9931
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.042904 0.989399 121
## 2 max f2 0.042904 0.987306 121
## 3 max f0point5 0.055867 0.995702 119
## 4 max accuracy 0.055867 0.999698 119
## 5 max precision 0.885481 1.000000 0
## 6 max recall 0.018184 1.000000 153
## 7 max specificity 0.885481 1.000000 0
## 8 max absolute_mcc 0.042904 0.989252 121
## 9 max min_per_class_accuracy 0.018184 0.993666 153
## 10 max mean_per_class_accuracy 0.018184 0.996833 153
## 11 max tns 0.885481 9789.000000 0
## 12 max fns 0.885481 141.000000 0
## 13 max fps 0.003120 9789.000000 399
## 14 max tps 0.018184 142.000000 153
## 15 max tnr 0.885481 1.000000 0
## 16 max fnr 0.885481 0.992958 0
## 17 max fpr 0.003120 1.000000 399
## 18 max tpr 0.018184 1.000000 153
##
## 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: 0.00177014
## RMSE: 0.04207304
## LogLoss: 0.01214736
## Mean Per-Class Error: 0.0141844
## AUC: 0.9945695
## AUCPR: 0.982806
## Gini: 0.989139
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## FALSE TRUE Error Rate
## FALSE 8536 0 0.000000 =0/8536
## TRUE 4 137 0.028369 =4/141
## Totals 8540 137 0.000461 =4/8677
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.627327 0.985612 124
## 2 max f2 0.627327 0.977175 124
## 3 max f0point5 0.627327 0.994194 124
## 4 max accuracy 0.627327 0.999539 124
## 5 max precision 0.912168 1.000000 0
## 6 max recall 0.004021 1.000000 343
## 7 max specificity 0.912168 1.000000 0
## 8 max absolute_mcc 0.627327 0.985483 124
## 9 max min_per_class_accuracy 0.015016 0.985816 170
## 10 max mean_per_class_accuracy 0.027117 0.988600 134
## 11 max tns 0.912168 8536.000000 0
## 12 max fns 0.912168 140.000000 0
## 13 max fps 0.003120 8536.000000 399
## 14 max tps 0.004021 141.000000 343
## 15 max tnr 0.912168 1.000000 0
## 16 max fnr 0.912168 0.992908 0
## 17 max fpr 0.003120 1.000000 399
## 18 max tpr 0.004021 1.000000 343
##
## 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: 0.0005149152
## RMSE: 0.02269174
## LogLoss: 0.004125167
## Mean Per-Class Error: 0.01798561
## AUC: 0.9937849
## AUCPR: 0.9742133
## Gini: 0.9875699
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## FALSE TRUE Error Rate
## FALSE 48017 0 0.000000 =0/48017
## TRUE 25 670 0.035971 =25/695
## Totals 48042 670 0.000513 =25/48712
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.957948 0.981685 140
## 2 max f2 0.957948 0.971014 140
## 3 max f0point5 0.957948 0.992593 140
## 4 max accuracy 0.957948 0.999487 140
## 5 max precision 0.999450 1.000000 0
## 6 max recall 0.000264 1.000000 395
## 7 max specificity 0.999450 1.000000 0
## 8 max absolute_mcc 0.957948 0.981594 140
## 9 max min_per_class_accuracy 0.002879 0.976050 273
## 10 max mean_per_class_accuracy 0.957948 0.982014 140
## 11 max tns 0.999450 48017.000000 0
## 12 max fns 0.999450 694.000000 0
## 13 max fps 0.000221 48017.000000 399
## 14 max tps 0.000264 695.000000 395
## 15 max tnr 0.999450 1.000000 0
## 16 max fnr 0.999450 0.998561 0
## 17 max fpr 0.000221 1.000000 399
## 18 max tpr 0.000264 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 0.999487 0.000072 0.999385 0.999484 0.999492 0.999588
## auc 0.994158 0.004146 0.988600 0.993990 0.992179 0.999463
## err 0.000513 0.000072 0.000615 0.000516 0.000508 0.000412
## err_count 5.000000 0.707107 6.000000 5.000000 5.000000 4.000000
## f0point5 0.992560 0.001162 0.990991 0.992424 0.993548 0.993837
## cv_5_valid
## accuracy 0.999485
## auc 0.996560
## err 0.000515
## err_count 5.000000
## f0point5 0.992000
##
## ---
## mean sd cv_1_valid cv_2_valid cv_3_valid
## precision 1.000000 0.000000 1.000000 1.000000 1.000000
## r2 0.963301 0.005646 0.955734 0.962662 0.967950
## recall 0.963895 0.005475 0.956522 0.963235 0.968553
## residual_deviance 79.477200 15.236052 100.412210 78.300640 81.349570
## rmse 0.022628 0.001624 0.024838 0.022727 0.022559
## specificity 1.000000 0.000000 1.000000 1.000000 1.000000
## cv_4_valid cv_5_valid
## precision 1.000000 1.000000
## r2 0.969656 0.960502
## recall 0.969925 0.961240
## residual_deviance 57.449340 79.874245
## rmse 0.020255 0.022763
## specificity 1.000000 1.000000best_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: 19,130 × 24
## predict FALSE. TRUE. expedition_id member_id peak_id peak_name year season
## <fct> <dbl> <dbl> <chr> <chr> <chr> <chr> <dbl> <chr>
## 1 FALSE 0.994 0.00552 AMAD78301 AMAD7830… AMAD Ama Dabl… 1978 Autumn
## 2 FALSE 0.995 0.00487 AMAD79101 AMAD7910… AMAD Ama Dabl… 1979 Spring
## 3 FALSE 0.993 0.00682 AMAD79101 AMAD7910… AMAD Ama Dabl… 1979 Spring
## 4 FALSE 0.987 0.0135 AMAD79101 AMAD7910… AMAD Ama Dabl… 1979 Spring
## 5 FALSE 0.993 0.00657 AMAD79101 AMAD7910… AMAD Ama Dabl… 1979 Spring
## 6 FALSE 0.996 0.00392 AMAD79101 AMAD7910… AMAD Ama Dabl… 1979 Spring
## 7 FALSE 0.995 0.00470 AMAD79101 AMAD7910… AMAD Ama Dabl… 1979 Spring
## 8 FALSE 0.996 0.00394 AMAD79101 AMAD7910… AMAD Ama Dabl… 1979 Spring
## 9 FALSE 0.992 0.00758 AMAD79101 AMAD7910… AMAD Ama Dabl… 1979 Spring
## 10 FALSE 0.996 0.00402 AMAD79301 AMAD7930… AMAD Ama Dabl… 1979 Autumn
## # ℹ 19,120 more rows
## # ℹ 15 more variables: sex <chr>, age <dbl>, citizenship <chr>,
## # expedition_role <chr>, hired <lgl>, highpoint_metres <dbl>, success <lgl>,
## # solo <lgl>, oxygen_used <lgl>, died <lgl>, death_cause <chr>,
## # death_height_metres <dbl>, injured <lgl>, injury_type <chr>,
## # injury_height_metres <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_21_20240430_114612"
##
## $model$type
## [1] "Key<Model>"
##
## $model$URL
## [1] "/3/Models/StackedEnsemble_BestOfFamily_1_AutoML_21_20240430_114612"
##
##
## $model_checksum
## [1] "2571579368806993928"
##
## $frame
## $frame$name
## [1] "test_tbl_sid_9611_3"
##
##
## $frame_checksum
## [1] "678340420273909232"
##
## $description
## NULL
##
## $scoring_time
## [1] 1.714492e+12
##
## $predictions
## NULL
##
## $MSE
## [1] 0.001620983
##
## $RMSE
## [1] 0.04026143
##
## $nobs
## [1] 19130
##
## $custom_metric_name
## NULL
##
## $custom_metric_value
## [1] 0
##
## $r2
## [1] 0.8835062
##
## $logloss
## [1] 0.01133142
##
## $AUC
## [1] 0.9940374
##
## $pr_auc
## [1] 0.9850477
##
## $Gini
## [1] 0.9880747
##
## $mean_per_class_error
## [1] 0.01666667
##
## $domain
## [1] "FALSE" "TRUE"
##
## $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
## FALSE TRUE Error Rate
## FALSE 18860 0 0.0000 = 0 / 18,860
## TRUE 9 261 0.0333 = 9 / 270
## Totals 18869 261 0.0005 = 9 / 19,130
##
##
## $thresholds_and_metric_scores
## Metrics for Thresholds: Binomial metrics as a function of classification thresholds
## threshold f1 f2 f0point5 accuracy precision recall specificity
## 1 0.915912 0.007380 0.004625 0.018248 0.985938 1.000000 0.003704 1.000000
## 2 0.897616 0.014706 0.009242 0.035971 0.985991 1.000000 0.007407 1.000000
## 3 0.889059 0.021978 0.013850 0.053191 0.986043 1.000000 0.011111 1.000000
## 4 0.888402 0.029197 0.018450 0.069930 0.986095 1.000000 0.014815 1.000000
## 5 0.887458 0.036364 0.023041 0.086207 0.986147 1.000000 0.018519 1.000000
## absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns fns fps tps
## 1 0.060429 0.003704 0.501852 18860 269 0 1
## 2 0.085461 0.007407 0.503704 18860 268 0 2
## 3 0.104671 0.011111 0.505556 18860 267 0 3
## 4 0.120867 0.014815 0.507407 18860 266 0 4
## 5 0.135137 0.018519 0.509259 18860 265 0 5
## tnr fnr fpr tpr idx
## 1 1.000000 0.996296 0.000000 0.003704 0
## 2 1.000000 0.992593 0.000000 0.007407 1
## 3 1.000000 0.988889 0.000000 0.011111 2
## 4 1.000000 0.985185 0.000000 0.014815 3
## 5 1.000000 0.981481 0.000000 0.018519 4
##
## ---
## threshold f1 f2 f0point5 accuracy precision recall
## 395 0.003238 0.029462 0.070537 0.018619 0.070099 0.014951 1.000000
## 396 0.003215 0.029026 0.069537 0.018341 0.055724 0.014727 1.000000
## 397 0.003190 0.028584 0.068521 0.018058 0.040669 0.014499 1.000000
## 398 0.003169 0.028284 0.067832 0.017867 0.030214 0.014345 1.000000
## 399 0.003153 0.028042 0.067275 0.017712 0.021589 0.014220 1.000000
## 400 0.003137 0.027835 0.066799 0.017580 0.014114 0.014114 1.000000
## specificity absolute_mcc min_per_class_accuracy mean_per_class_accuracy
## 395 0.056787 0.029138 0.056787 0.528393
## 396 0.042206 0.024931 0.042206 0.521103
## 397 0.026935 0.019762 0.026935 0.513468
## 398 0.016331 0.015306 0.016331 0.508165
## 399 0.007582 0.010384 0.007582 0.503791
## 400 0.000000 0.000000 0.000000 0.500000
## tns fns fps tps tnr fnr fpr tpr idx
## 395 1071 0 17789 270 0.056787 0.000000 0.943213 1.000000 394
## 396 796 0 18064 270 0.042206 0.000000 0.957794 1.000000 395
## 397 508 0 18352 270 0.026935 0.000000 0.973065 1.000000 396
## 398 308 0 18552 270 0.016331 0.000000 0.983669 1.000000 397
## 399 143 0 18717 270 0.007582 0.000000 0.992418 1.000000 398
## 400 0 0 18860 270 0.000000 0.000000 1.000000 1.000000 399
##
## $max_criteria_and_metric_scores
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.586446 0.983051 176
## 2 max f2 0.031377 0.980755 181
## 3 max f0point5 0.586446 0.993151 176
## 4 max accuracy 0.586446 0.999530 176
## 5 max precision 0.915912 1.000000 0
## 6 max recall 0.003644 1.000000 372
## 7 max specificity 0.915912 1.000000 0
## 8 max absolute_mcc 0.586446 0.982958 176
## 9 max min_per_class_accuracy 0.017392 0.985185 211
## 10 max mean_per_class_accuracy 0.031377 0.990582 181
## 11 max tns 0.915912 18860.000000 0
## 12 max fns 0.915912 269.000000 0
## 13 max fps 0.003137 18860.000000 399
## 14 max tps 0.003644 270.000000 372
## 15 max tnr 0.915912 1.000000 0
## 16 max fnr 0.915912 0.996296 0
## 17 max fpr 0.003137 1.000000 399
## 18 max tpr 0.003644 1.000000 372
##
## $gains_lift_table
## Gains/Lift Table: Avg response rate: 1.41 %, avg score: 1.45 %
## group cumulative_data_fraction lower_threshold lift cumulative_lift
## 1 1 0.01003659 0.681112 70.851852 70.851852
## 2 2 0.02002091 0.017363 27.450456 49.207814
## 3 3 0.03037114 0.012151 0.000000 32.438197
## 4 4 0.04004182 0.010209 0.000000 24.603907
## 5 5 0.05013068 0.008637 0.000000 19.652338
## 6 6 0.10000000 0.006689 0.074268 9.888889
## 7 7 0.15018296 0.005893 0.000000 6.584561
## 8 8 0.20010455 0.005731 0.000000 4.941861
## 9 9 0.30000000 0.005532 0.000000 3.296296
## 10 10 0.40000000 0.004474 0.000000 2.472222
## 11 11 0.50000000 0.003958 0.074074 1.992593
## 12 12 0.60000000 0.003690 0.000000 1.660494
## 13 13 0.70000000 0.003533 0.037037 1.428571
## 14 14 0.80000000 0.003410 0.000000 1.250000
## 15 15 0.90000000 0.003290 0.000000 1.111111
## 16 16 1.00000000 0.003097 0.000000 1.000000
## response_rate score cumulative_response_rate cumulative_score
## 1 1.000000 0.738970 1.000000 0.738970
## 2 0.387435 0.252440 0.694517 0.496340
## 3 0.000000 0.014398 0.457831 0.332098
## 4 0.000000 0.011022 0.347258 0.254554
## 5 0.000000 0.009252 0.277372 0.205187
## 6 0.001048 0.007472 0.139571 0.106588
## 7 0.000000 0.006193 0.092934 0.073041
## 8 0.000000 0.005802 0.069749 0.056267
## 9 0.000000 0.005619 0.046524 0.039402
## 10 0.000000 0.005066 0.034893 0.030818
## 11 0.001045 0.004189 0.028123 0.025492
## 12 0.000000 0.003811 0.023436 0.021878
## 13 0.000523 0.003610 0.020163 0.019269
## 14 0.000000 0.003470 0.017642 0.017294
## 15 0.000000 0.003354 0.015682 0.015745
## 16 0.000000 0.003214 0.014114 0.014492
## capture_rate cumulative_capture_rate gain cumulative_gain
## 1 0.711111 0.711111 6985.185185 6985.185185
## 2 0.274074 0.985185 2645.045569 4820.781356
## 3 0.000000 0.985185 -100.000000 3143.819723
## 4 0.000000 0.985185 -100.000000 2360.390678
## 5 0.000000 0.985185 -100.000000 1865.233847
## 6 0.003704 0.988889 -92.573181 888.888889
## 7 0.000000 0.988889 -100.000000 558.456124
## 8 0.000000 0.988889 -100.000000 394.186114
## 9 0.000000 0.988889 -100.000000 229.629630
## 10 0.000000 0.988889 -100.000000 147.222222
## 11 0.007407 0.996296 -92.592593 99.259259
## 12 0.000000 0.996296 -100.000000 66.049383
## 13 0.003704 1.000000 -96.296296 42.857143
## 14 0.000000 1.000000 -100.000000 25.000000
## 15 0.000000 1.000000 -100.000000 11.111111
## 16 0.000000 1.000000 -100.000000 0.000000
## kolmogorov_smirnov
## 1 0.711111
## 2 0.978982
## 3 0.968483
## 4 0.958674
## 5 0.948441
## 6 0.901614
## 7 0.850713
## 8 0.800077
## 9 0.698751
## 10 0.597319
## 11 0.503401
## 12 0.401970
## 13 0.304295
## 14 0.202863
## 15 0.101432
## 16 0.000000
##
## $residual_deviance
## [1] 433.5403
##
## $null_deviance
## [1] 2836.923
##
## $AIC
## [1] 439.5403
##
## $loglikelihood
## [1] 0
##
## $null_degrees_of_freedom
## [1] 19129
##
## $residual_degrees_of_freedom
## [1] 19127h2o.auc(best_model)## [1] 0.9999169h2o.confusionMatrix(performance_h2o)## Confusion Matrix (vertical: actual; across: predicted) for max f1 @ threshold = 0.586446376329621:
## FALSE TRUE Error Rate
## FALSE 18860 0 0.000000 =0/18860
## TRUE 9 261 0.033333 =9/270
## Totals 18869 261 0.000470 =9/19130h2o.metric(performance_h2o) %>% as_tibble() %>% filter(threshold %>% between(0.98, 0.99))## # A tibble: 0 × 20
## # ℹ 20 variables: threshold <dbl>, f1 <dbl>, f2 <dbl>, f0point5 <dbl>,
## # accuracy <dbl>, precision <dbl>, recall <dbl>, specificity <dbl>,
## # 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>