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.csv") %>%
# h2o requires all variables to be either numeric or factors
mutate(across(where(is.character), factor))set.seed(1234)
data_split <- initial_split(data, strata = "accreditation")
train_tbl <- training(data_split)
test_tbl <- testing(data_split)recipe_obj <- recipe(accreditation ~ ., 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: 2 days 24 minutes
## H2O cluster timezone: America/New_York
## H2O data parsing timezone: UTC
## H2O cluster version: 3.44.0.3
## H2O cluster version age: 11 months
## H2O cluster name: H2O_started_from_R_jordanlanowy_fhp551
## H2O cluster total nodes: 1
## H2O cluster total memory: 1.44 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 (11 months) 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 <- "accreditation"
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%
|
|=== | 4%
## 13:24:45.800: 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%
|
|============= | 18%
|
|================= | 25%
|
|====================== | 32%
|
|=========================== | 39%
|
|================================ | 46%
|
|===================================== | 53%
|
|========================================== | 60%
|
|============================================== | 66%
|
|=================================================== | 73%
|
|======================================================== | 80%
|
|============================================================= | 87%
|
|======================================================================| 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_AllModels_1_AutoML_15_20241121_132445 0.9371310 0.3047063
## 2 StackedEnsemble_BestOfFamily_1_AutoML_15_20241121_132445 0.9366797 0.3060339
## 3 XGBoost_3_AutoML_15_20241121_132445 0.9353839 0.3133164
## 4 GLM_1_AutoML_15_20241121_132445 0.9336554 0.3132786
## 5 XGBoost_2_AutoML_15_20241121_132445 0.9237338 0.3484276
## 6 XGBoost_1_AutoML_15_20241121_132445 0.9200615 0.3491579
## aucpr mean_per_class_error rmse mse
## 1 0.9594277 0.1095772 0.3045837 0.09277121
## 2 0.9584524 0.1168927 0.3042431 0.09256386
## 3 0.9577012 0.1218670 0.3076886 0.09467225
## 4 0.9575205 0.1210038 0.3083793 0.09509777
## 5 0.9472094 0.1242517 0.3229280 0.10428252
## 6 0.9484182 0.1280523 0.3255031 0.10595228
##
## [12 rows x 7 columns]models_h2o@leader## Model Details:
## ==============
##
## H2OBinomialModel: stackedensemble
## Model ID: StackedEnsemble_AllModels_1_AutoML_15_20241121_132445
## Model Summary for Stacked Ensemble:
## key value
## 1 Stacking strategy cross_validation
## 2 Number of base models (used / total) 4/10
## 3 # GBM base models (used / total) 1/4
## 4 # XGBoost base models (used / total) 1/3
## 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: 0.03770232
## RMSE: 0.1941709
## LogLoss: 0.1437038
## Mean Per-Class Error: 0.0459837
## AUC: 0.993834
## AUCPR: 0.995356
## Gini: 0.987668
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Accredited Unaccredited Error Rate
## Accredited 1004 81 0.074654 =81/1085
## Unaccredited 25 1419 0.017313 =25/1444
## Totals 1029 1500 0.041914 =106/2529
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.317940 0.963995 237
## 2 max f2 0.249712 0.981583 257
## 3 max f0point5 0.612197 0.968843 154
## 4 max accuracy 0.317940 0.958086 237
## 5 max precision 0.999674 1.000000 0
## 6 max recall 0.174263 1.000000 285
## 7 max specificity 0.999674 1.000000 0
## 8 max absolute_mcc 0.317940 0.914843 237
## 9 max min_per_class_accuracy 0.447828 0.949309 202
## 10 max mean_per_class_accuracy 0.317940 0.954016 237
## 11 max tns 0.999674 1085.000000 0
## 12 max fns 0.999674 1307.000000 0
## 13 max fps 0.005000 1085.000000 399
## 14 max tps 0.174263 1444.000000 285
## 15 max tnr 0.999674 1.000000 0
## 16 max fnr 0.999674 0.905125 0
## 17 max fpr 0.005000 1.000000 399
## 18 max tpr 0.174263 1.000000 285
##
## 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: 0.08736751
## RMSE: 0.29558
## LogLoss: 0.2921939
## Mean Per-Class Error: 0.1110858
## AUC: 0.9472648
## AUCPR: 0.9653165
## Gini: 0.8945296
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Accredited Unaccredited Error Rate
## Accredited 162 28 0.147368 =28/190
## Unaccredited 19 235 0.074803 =19/254
## Totals 181 263 0.105856 =47/444
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.590911 0.909091 225
## 2 max f2 0.234776 0.928189 254
## 3 max f0point5 0.879082 0.927788 172
## 4 max accuracy 0.614111 0.894144 223
## 5 max precision 0.999863 1.000000 0
## 6 max recall 0.043830 1.000000 387
## 7 max specificity 0.999863 1.000000 0
## 8 max absolute_mcc 0.614111 0.783198 223
## 9 max min_per_class_accuracy 0.755629 0.885827 207
## 10 max mean_per_class_accuracy 0.755629 0.890282 207
## 11 max tns 0.999863 190.000000 0
## 12 max fns 0.999863 252.000000 0
## 13 max fps 0.007441 190.000000 399
## 14 max tps 0.043830 254.000000 387
## 15 max tnr 0.999863 1.000000 0
## 16 max fnr 0.999863 0.992126 0
## 17 max fpr 0.007441 1.000000 399
## 18 max tpr 0.043830 1.000000 387
##
## 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: 0.08671589
## RMSE: 0.2944756
## LogLoss: 0.290198
## Mean Per-Class Error: 0.1057492
## AUC: 0.941625
## AUCPR: 0.9623053
## Gini: 0.8832499
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Accredited Unaccredited Error Rate
## Accredited 978 107 0.098618 =107/1085
## Unaccredited 163 1281 0.112881 =163/1444
## Totals 1141 1388 0.106762 =270/2529
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.446059 0.904661 200
## 2 max f2 0.170718 0.909633 298
## 3 max f0point5 0.719219 0.924312 143
## 4 max accuracy 0.519811 0.894029 187
## 5 max precision 0.999726 1.000000 0
## 6 max recall 0.026721 1.000000 392
## 7 max specificity 0.999726 1.000000 0
## 8 max absolute_mcc 0.519811 0.787407 187
## 9 max min_per_class_accuracy 0.413299 0.891244 208
## 10 max mean_per_class_accuracy 0.519811 0.896547 187
## 11 max tns 0.999726 1085.000000 0
## 12 max fns 0.999726 1380.000000 0
## 13 max fps 0.004505 1085.000000 399
## 14 max tps 0.026721 1444.000000 392
## 15 max tnr 0.999726 1.000000 0
## 16 max fnr 0.999726 0.955679 0
## 17 max fpr 0.004505 1.000000 399
## 18 max tpr 0.026721 1.000000 392
##
## 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 0.897202 0.022957 0.931481 0.910506 0.882235 0.881092
## auc 0.942244 0.010531 0.952546 0.953077 0.940730 0.928876
## err 0.102798 0.022957 0.068519 0.089494 0.117764 0.118908
## err_count 51.600000 9.989995 37.000000 46.000000 59.000000 61.000000
## f0point5 0.922940 0.014994 0.947622 0.922598 0.908767 0.913534
## cv_5_valid
## accuracy 0.880694
## auc 0.935989
## err 0.119306
## err_count 55.000000
## f0point5 0.922179
##
## ---
## mean sd cv_1_valid cv_2_valid cv_3_valid
## precision 0.933503 0.014724 0.954861 0.923875 0.917266
## r2 0.643981 0.054988 0.721783 0.680706 0.619601
## recall 0.883759 0.033342 0.919732 0.917526 0.876289
## residual_deviance 293.146200 23.289720 268.940920 278.529720 302.449900
## rmse 0.294251 0.022277 0.262206 0.280047 0.304325
## specificity 0.915413 0.021191 0.946058 0.901345 0.890476
## cv_4_valid cv_5_valid
## precision 0.931034 0.940476
## r2 0.603117 0.594699
## recall 0.849650 0.855596
## residual_deviance 328.495180 287.315340
## rmse 0.312903 0.311772
## specificity 0.920705 0.918478?h2o.getModel
?h2o.saveModel
?h2o.loadModel
h2o.getModel("StackedEnsemble_AllModels_1_AutoML_13_20241121_130635") %>%
h2o.saveModel("h2o_models/")## [1] "/Users/jordanlanowy/Desktop/PSU_DAT3100/11_module13/h2o_models/StackedEnsemble_AllModels_1_AutoML_13_20241121_130635"best_model <- h2o.loadModel("h2o_models/StackedEnsemble_AllModels_1_AutoML_13_20241121_130635 /StackedEnsemble_AllModels_1_AutoML_13_20241121_130635")
best_model## Model Details:
## ==============
##
## H2OBinomialModel: stackedensemble
## Model ID: StackedEnsemble_AllModels_1_AutoML_13_20241121_130635
## Model Summary for Stacked Ensemble:
## key value
## 1 Stacking strategy cross_validation
## 2 Number of base models (used / total) 4/10
## 3 # GBM base models (used / total) 1/4
## 4 # XGBoost base models (used / total) 1/3
## 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: 0.03770232
## RMSE: 0.1941709
## LogLoss: 0.1437038
## Mean Per-Class Error: 0.0459837
## AUC: 0.993834
## AUCPR: 0.995356
## Gini: 0.987668
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Accredited Unaccredited Error Rate
## Accredited 1004 81 0.074654 =81/1085
## Unaccredited 25 1419 0.017313 =25/1444
## Totals 1029 1500 0.041914 =106/2529
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.317940 0.963995 237
## 2 max f2 0.249712 0.981583 257
## 3 max f0point5 0.612197 0.968843 154
## 4 max accuracy 0.317940 0.958086 237
## 5 max precision 0.999674 1.000000 0
## 6 max recall 0.174263 1.000000 285
## 7 max specificity 0.999674 1.000000 0
## 8 max absolute_mcc 0.317940 0.914843 237
## 9 max min_per_class_accuracy 0.447828 0.949309 202
## 10 max mean_per_class_accuracy 0.317940 0.954016 237
## 11 max tns 0.999674 1085.000000 0
## 12 max fns 0.999674 1307.000000 0
## 13 max fps 0.005000 1085.000000 399
## 14 max tps 0.174263 1444.000000 285
## 15 max tnr 0.999674 1.000000 0
## 16 max fnr 0.999674 0.905125 0
## 17 max fpr 0.005000 1.000000 399
## 18 max tpr 0.174263 1.000000 285
##
## 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: 0.08736751
## RMSE: 0.29558
## LogLoss: 0.2921939
## Mean Per-Class Error: 0.1110858
## AUC: 0.9472648
## AUCPR: 0.9653165
## Gini: 0.8945296
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Accredited Unaccredited Error Rate
## Accredited 162 28 0.147368 =28/190
## Unaccredited 19 235 0.074803 =19/254
## Totals 181 263 0.105856 =47/444
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.590911 0.909091 225
## 2 max f2 0.234776 0.928189 254
## 3 max f0point5 0.879082 0.927788 172
## 4 max accuracy 0.614111 0.894144 223
## 5 max precision 0.999863 1.000000 0
## 6 max recall 0.043830 1.000000 387
## 7 max specificity 0.999863 1.000000 0
## 8 max absolute_mcc 0.614111 0.783198 223
## 9 max min_per_class_accuracy 0.755629 0.885827 207
## 10 max mean_per_class_accuracy 0.755629 0.890282 207
## 11 max tns 0.999863 190.000000 0
## 12 max fns 0.999863 252.000000 0
## 13 max fps 0.007441 190.000000 399
## 14 max tps 0.043830 254.000000 387
## 15 max tnr 0.999863 1.000000 0
## 16 max fnr 0.999863 0.992126 0
## 17 max fpr 0.007441 1.000000 399
## 18 max tpr 0.043830 1.000000 387
##
## 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: 0.08671589
## RMSE: 0.2944756
## LogLoss: 0.290198
## Mean Per-Class Error: 0.1057492
## AUC: 0.941625
## AUCPR: 0.9623053
## Gini: 0.8832499
##
## Confusion Matrix (vertical: actual; across: predicted) for F1-optimal threshold:
## Accredited Unaccredited Error Rate
## Accredited 978 107 0.098618 =107/1085
## Unaccredited 163 1281 0.112881 =163/1444
## Totals 1141 1388 0.106762 =270/2529
##
## Maximum Metrics: Maximum metrics at their respective thresholds
## metric threshold value idx
## 1 max f1 0.446059 0.904661 200
## 2 max f2 0.170718 0.909633 298
## 3 max f0point5 0.719219 0.924312 143
## 4 max accuracy 0.519811 0.894029 187
## 5 max precision 0.999726 1.000000 0
## 6 max recall 0.026721 1.000000 392
## 7 max specificity 0.999726 1.000000 0
## 8 max absolute_mcc 0.519811 0.787407 187
## 9 max min_per_class_accuracy 0.413299 0.891244 208
## 10 max mean_per_class_accuracy 0.519811 0.896547 187
## 11 max tns 0.999726 1085.000000 0
## 12 max fns 0.999726 1380.000000 0
## 13 max fps 0.004505 1085.000000 399
## 14 max tps 0.026721 1444.000000 392
## 15 max tnr 0.999726 1.000000 0
## 16 max fnr 0.999726 0.955679 0
## 17 max fpr 0.004505 1.000000 399
## 18 max tpr 0.026721 1.000000 392
##
## 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 0.897202 0.022957 0.931481 0.910506 0.882235 0.881092
## auc 0.942244 0.010531 0.952546 0.953077 0.940730 0.928876
## err 0.102798 0.022957 0.068519 0.089494 0.117764 0.118908
## err_count 51.600000 9.989995 37.000000 46.000000 59.000000 61.000000
## f0point5 0.922940 0.014994 0.947622 0.922598 0.908767 0.913534
## cv_5_valid
## accuracy 0.880694
## auc 0.935989
## err 0.119306
## err_count 55.000000
## f0point5 0.922179
##
## ---
## mean sd cv_1_valid cv_2_valid cv_3_valid
## precision 0.933503 0.014724 0.954861 0.923875 0.917266
## r2 0.643981 0.054988 0.721783 0.680706 0.619601
## recall 0.883759 0.033342 0.919732 0.917526 0.876289
## residual_deviance 293.146200 23.289720 268.940920 278.529720 302.449900
## rmse 0.294251 0.022277 0.262206 0.280047 0.304325
## specificity 0.915413 0.021191 0.946058 0.901345 0.890476
## cv_4_valid cv_5_valid
## precision 0.931034 0.940476
## r2 0.603117 0.594699
## recall 0.849650 0.855596
## residual_deviance 328.495180 287.315340
## rmse 0.312903 0.311772
## specificity 0.920705 0.918478predictions <- h2o.predict(best_model, newdata = test_h2o)##
|
| | 0%
|
|======================================================================| 100%## Warning in doTryCatch(return(expr), name, parentenv, handler): Test/Validation
## dataset column 'name_of_museum' has levels not trained on: ["1066 Battle Of
## Hastings, Battle Abbey And Battlefield", "13th/18th Royal Hussars Museum",
## "1620s House And Garden", "Abbotsford", "Aberaeron Toy Museum", "Aberdeen
## Science Centre", "Aberdeenshire Farming Museum", "Abergavenny Museum",
## "Abertillery And District Museum", "Abriachan Museum", ...973 not listed...,
## "World Of Glass", "Wrexham County Borough Museum", "Wymondham Abbey", "Yelde
## Hall Museum", "Yelverton Paperweight Centre", "York City Art Gallery", "York
## Cold War Bunker", "Yorkshire Museum Of Farming", "Yorkshire Sculpture Park",
## "Young Gallery Salisbury"]## Warning in doTryCatch(return(expr), name, parentenv, handler): Test/Validation
## dataset column 'village_town_or_city' has levels not trained on: ["Aberaeron",
## "Abriachan", "Alexandria", "Antrim", "Appleby Magna", "Argyll", "Arisaig",
## "Ash", "Ashby-de-la-Zouch", "Ballater", ...243 not listed..., "nr Okehampton",
## "nr Prestonpans", "nr Skipton", "nr Spilsby", "nr Wakefield", "nr Westbury",
## "nr. Cheltenham", "nr. Kilwinning", "nr. Rugeley", "nr. Yeovil"]## Warning in doTryCatch(return(expr), name, parentenv, handler): Test/Validation
## dataset column 'postcode' has levels not trained on: ["AB10 1JX", "AB11 6EQ",
## "AB15 7XH", "AB24 3EE", "AB32 6RX", "AB35 5TB", "AB41 7GX", "AB41 7PD", "AB42
## 2UP", "AB42 5FQ", ...876 not listed..., "YO61 1DP", "YO61 4AD", "YO62 5LJ",
## "ZE1 0EL", "ZE1 0UL", "ZE2 9AY", "ZE2 9DJ", "ZE2 9HL", "ZE2 9QD", "ZE2 9SB"]## Warning in doTryCatch(return(expr), name, parentenv, handler): Test/Validation
## dataset column 'admin_area' has levels not trained on: ["/England/South East
## (English Region)/Bracknell Forest (English UA)", "/England/South West (English
## Region)/Isles of Scilly (English UA)"]## Warning in doTryCatch(return(expr), name, parentenv, handler): Test/Validation
## dataset column 'subject_matter' has levels not trained on: ["Communications"]## Warning in doTryCatch(return(expr), name, parentenv, handler): Test/Validation
## dataset column 'year_opened' has levels not trained on: ["1660:1660",
## "1707:1707", "1750:1759", "1761:1761", "1807:1807", "1816:1816", "1818:1818",
## "1823:1823", "1843:1843", "1854:1854", ...15 not listed..., "1960:1979",
## "1960:1984", "1960:1997", "1960:2000", "1974:1979", "1983:1985", "1989:2004",
## "1995:1996", "1996:1998", "2000:2005"]## Warning in doTryCatch(return(expr), name, parentenv, handler): Test/Validation
## dataset column 'year_closed' has levels not trained on: ["1936:2017",
## "1965:2017", "1971:1999", "1978:1978", "1979:1981", "1981:2017", "1984:1985",
## "1984:2004", "1985:2002", "1985:2008", "1985:2015", "1990:2000", "1991:2017",
## "2006:2017", "2007:2017", "2011:2017", "2017:2018"]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: 993 × 29
## predict Accredited Unaccredited X name_of_museum village_town_or_city
## <fct> <dbl> <dbl> <int> <fct> <fct>
## 1 Unaccredit… 0.00199 0.998 1 The Woodland … nr Westbury
## 2 Unaccredit… 0.126 0.874 3 Jet Age Museum Cheltenham
## 3 Unaccredit… 0.306 0.694 6 Prestongrange… nr Prestonpans
## 4 Accredited 0.955 0.0451 7 National Muse… Edinburgh
## 5 Accredited 0.782 0.218 9 Baird Institu… Cumnock
## 6 Accredited 0.853 0.147 11 Taigh-tasgaid… Kildonan
## 7 Unaccredit… 0.0398 0.960 15 Clayton Hall … Manchester
## 8 Accredited 0.930 0.0700 20 National War … Edinburgh
## 9 Unaccredit… 0.00593 0.994 21 Woodlands Art… London
## 10 Unaccredit… 0.000306 1.00 25 London Gas Mu… London
## # ℹ 983 more rows
## # ℹ 23 more variables: postcode <fct>, latitude <dbl>, longitude <dbl>,
## # admin_area <fct>, accreditation <fct>, governance <fct>, size <fct>,
## # size_provenance <fct>, subject_matter <fct>, year_opened <fct>,
## # year_closed <fct>, primary_provenance_of_data <fct>,
## # area_deprivation_index <int>, area_deprivation_index_crime <int>,
## # area_deprivation_index_education <int>, …?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_AllModels_1_AutoML_13_20241121_130635"
##
## $model$type
## [1] "Key<Model>"
##
## $model$URL
## [1] "/3/Models/StackedEnsemble_AllModels_1_AutoML_13_20241121_130635"
##
##
## $model_checksum
## [1] "-7071169776820997318"
##
## $frame
## $frame$name
## [1] "test_tbl_sid_b0ec_3"
##
##
## $frame_checksum
## [1] "2942151053878311765"
##
## $description
## NULL
##
## $scoring_time
## [1] 1.732214e+12
##
## $predictions
## NULL
##
## $MSE
## [1] 0.09277121
##
## $RMSE
## [1] 0.3045837
##
## $nobs
## [1] 993
##
## $custom_metric_name
## NULL
##
## $custom_metric_value
## [1] 0
##
## $r2
## [1] 0.6212793
##
## $logloss
## [1] 0.3047063
##
## $AUC
## [1] 0.937131
##
## $pr_auc
## [1] 0.9594277
##
## $Gini
## [1] 0.874262
##
## $mean_per_class_error
## [1] 0.1095772
##
## $domain
## [1] "Accredited" "Unaccredited"
##
## $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
## Accredited Unaccredited Error Rate
## Accredited 395 31 0.0728 = 31 / 426
## Unaccredited 83 484 0.1464 = 83 / 567
## Totals 478 515 0.1148 = 114 / 993
##
##
## $thresholds_and_metric_scores
## Metrics for Thresholds: Binomial metrics as a function of classification thresholds
## threshold f1 f2 f0point5 accuracy precision recall specificity
## 1 0.999679 0.103679 0.067421 0.224313 0.460222 1.000000 0.054674 1.000000
## 2 0.999337 0.173913 0.116279 0.344828 0.483384 1.000000 0.095238 1.000000
## 3 0.999091 0.233645 0.160051 0.432526 0.504532 1.000000 0.132275 1.000000
## 4 0.998800 0.276596 0.192878 0.488722 0.520645 1.000000 0.160494 1.000000
## 5 0.998473 0.302395 0.213170 0.520082 0.530715 1.000000 0.178131 1.000000
## absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns fns fps tps
## 1 0.155599 0.054674 0.527337 426 536 0 31
## 2 0.207863 0.095238 0.547619 426 513 0 54
## 3 0.247755 0.132275 0.566138 426 492 0 75
## 4 0.275316 0.160494 0.580247 426 476 0 91
## 5 0.291670 0.178131 0.589065 426 466 0 101
## tnr fnr fpr tpr idx
## 1 1.000000 0.945326 0.000000 0.054674 0
## 2 1.000000 0.904762 0.000000 0.095238 1
## 3 1.000000 0.867725 0.000000 0.132275 2
## 4 1.000000 0.839506 0.000000 0.160494 3
## 5 1.000000 0.821869 0.000000 0.178131 4
##
## ---
## threshold f1 f2 f0point5 accuracy precision recall
## 395 0.015793 0.730200 0.871235 0.628464 0.578046 0.575051 1.000000
## 396 0.013890 0.729260 0.870700 0.627351 0.576032 0.573887 1.000000
## 397 0.011867 0.728792 0.870433 0.626796 0.575025 0.573306 1.000000
## 398 0.010070 0.728324 0.870166 0.626243 0.574018 0.572727 1.000000
## 399 0.007493 0.727856 0.869899 0.625690 0.573011 0.572149 1.000000
## 400 0.004727 0.726923 0.869365 0.624587 0.570997 0.570997 1.000000
## specificity absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns
## 395 0.016432 0.097207 0.016432 0.508216 7
## 396 0.011737 0.082072 0.011737 0.505869 5
## 397 0.009390 0.073370 0.009390 0.504695 4
## 398 0.007042 0.063508 0.007042 0.503521 3
## 399 0.004695 0.051828 0.004695 0.502347 2
## 400 0.000000 0.000000 0.000000 0.500000 0
## fns fps tps tnr fnr fpr tpr idx
## 395 0 419 567 0.016432 0.000000 0.983568 1.000000 394
## 396 0 421 567 0.011737 0.000000 0.988263 1.000000 395
## 397 0 422 567 0.009390 0.000000 0.990610 1.000000 396
## 398 0 423 567 0.007042 0.000000 0.992958 1.000000 397
## 399 0 424 567 0.004695 0.000000 0.995305 1.000000 398
## 400 0 426 567 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 0.666006 0.894640 162
## 2 max f2 0.150210 0.903614 297
## 3 max f0point5 0.752670 0.922985 143
## 4 max accuracy 0.666006 0.885196 162
## 5 max precision 0.999679 1.000000 0
## 6 max recall 0.024562 1.000000 391
## 7 max specificity 0.999679 1.000000 0
## 8 max absolute_mcc 0.666006 0.773471 162
## 9 max min_per_class_accuracy 0.436560 0.876543 191
## 10 max mean_per_class_accuracy 0.666006 0.890423 162
## 11 max tns 0.999679 426.000000 0
## 12 max fns 0.999679 536.000000 0
## 13 max fps 0.004727 426.000000 399
## 14 max tps 0.024562 567.000000 391
## 15 max tnr 0.999679 1.000000 0
## 16 max fnr 0.999679 0.945326 0
## 17 max fpr 0.004727 1.000000 399
## 18 max tpr 0.024562 1.000000 391
##
## $gains_lift_table
## Gains/Lift Table: Avg response rate: 57.10 %, avg score: 57.01 %
## group cumulative_data_fraction lower_threshold lift cumulative_lift
## 1 1 0.01007049 0.999738 1.751323 1.751323
## 2 2 0.02014099 0.999599 1.751323 1.751323
## 3 3 0.03021148 0.999542 1.751323 1.751323
## 4 4 0.04028197 0.999354 1.751323 1.751323
## 5 5 0.05035247 0.999247 1.751323 1.751323
## 6 6 0.10070493 0.998372 1.751323 1.751323
## 7 7 0.15005035 0.994811 1.751323 1.751323
## 8 8 0.20040282 0.989992 1.751323 1.751323
## 9 9 0.30010070 0.965191 1.698252 1.733692
## 10 10 0.39979859 0.897588 1.645182 1.711620
## 11 11 0.50050352 0.722766 1.436085 1.656180
## 12 12 0.60020141 0.276711 0.778366 1.510369
## 13 13 0.69989930 0.172740 0.336112 1.343101
## 14 14 0.79959718 0.108928 0.300732 1.213133
## 15 15 0.89929507 0.065965 0.194591 1.100215
## 16 16 1.00000000 0.004407 0.105079 1.000000
## response_rate score cumulative_response_rate cumulative_score
## 1 1.000000 0.999806 1.000000 0.999806
## 2 1.000000 0.999677 1.000000 0.999742
## 3 1.000000 0.999576 1.000000 0.999687
## 4 1.000000 0.999465 1.000000 0.999631
## 5 1.000000 0.999277 1.000000 0.999560
## 6 1.000000 0.998900 1.000000 0.999230
## 7 1.000000 0.996624 1.000000 0.998373
## 8 1.000000 0.992514 1.000000 0.996901
## 9 0.969697 0.979834 0.989933 0.991231
## 10 0.939394 0.937904 0.977330 0.977933
## 11 0.820000 0.825460 0.945674 0.947254
## 12 0.444444 0.482238 0.862416 0.870012
## 13 0.191919 0.214721 0.766906 0.776668
## 14 0.171717 0.136078 0.692695 0.696796
## 15 0.111111 0.087457 0.628219 0.629243
## 16 0.060000 0.041806 0.570997 0.570085
## capture_rate cumulative_capture_rate gain cumulative_gain
## 1 0.017637 0.017637 75.132275 75.132275
## 2 0.017637 0.035273 75.132275 75.132275
## 3 0.017637 0.052910 75.132275 75.132275
## 4 0.017637 0.070547 75.132275 75.132275
## 5 0.017637 0.088183 75.132275 75.132275
## 6 0.088183 0.176367 75.132275 75.132275
## 7 0.086420 0.262787 75.132275 75.132275
## 8 0.088183 0.350970 75.132275 75.132275
## 9 0.169312 0.520282 69.825236 73.369199
## 10 0.164021 0.684303 64.518198 71.162022
## 11 0.144621 0.828924 43.608466 65.618047
## 12 0.077601 0.906526 -22.163433 51.036895
## 13 0.033510 0.940035 -66.388755 34.310076
## 14 0.029982 0.970018 -69.926781 21.313289
## 15 0.019400 0.989418 -80.540858 10.021508
## 16 0.010582 1.000000 -89.492063 0.000000
## kolmogorov_smirnov
## 1 0.017637
## 2 0.035273
## 3 0.052910
## 4 0.070547
## 5 0.088183
## 6 0.176367
## 7 0.262787
## 8 0.350970
## 9 0.513240
## 10 0.663177
## 11 0.765544
## 12 0.714037
## 13 0.559754
## 14 0.397248
## 15 0.210075
## 16 0.000000
##
## $residual_deviance
## [1] 605.1466
##
## $null_deviance
## [1] 1356.501
##
## $AIC
## [1] 615.1466
##
## $loglikelihood
## [1] 0
##
## $null_degrees_of_freedom
## [1] 992
##
## $residual_degrees_of_freedom
## [1] 988h2o.auc(performance_h2o)## [1] 0.937131h2o.confusionMatrix(performance_h2o)## Confusion Matrix (vertical: actual; across: predicted) for max f1 @ threshold = 0.666005878614975:
## Accredited Unaccredited Error Rate
## Accredited 395 31 0.072770 =31/426
## Unaccredited 83 484 0.146384 =83/567
## Totals 478 515 0.114804 =114/993h2o.metric(performance_h2o)## Metrics for Thresholds: Binomial metrics as a function of classification thresholds
## threshold f1 f2 f0point5 accuracy precision recall specificity
## 1 0.999679 0.103679 0.067421 0.224313 0.460222 1.000000 0.054674 1.000000
## 2 0.999337 0.173913 0.116279 0.344828 0.483384 1.000000 0.095238 1.000000
## 3 0.999091 0.233645 0.160051 0.432526 0.504532 1.000000 0.132275 1.000000
## 4 0.998800 0.276596 0.192878 0.488722 0.520645 1.000000 0.160494 1.000000
## 5 0.998473 0.302395 0.213170 0.520082 0.530715 1.000000 0.178131 1.000000
## absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns fns fps tps
## 1 0.155599 0.054674 0.527337 426 536 0 31
## 2 0.207863 0.095238 0.547619 426 513 0 54
## 3 0.247755 0.132275 0.566138 426 492 0 75
## 4 0.275316 0.160494 0.580247 426 476 0 91
## 5 0.291670 0.178131 0.589065 426 466 0 101
## tnr fnr fpr tpr idx
## 1 1.000000 0.945326 0.000000 0.054674 0
## 2 1.000000 0.904762 0.000000 0.095238 1
## 3 1.000000 0.867725 0.000000 0.132275 2
## 4 1.000000 0.839506 0.000000 0.160494 3
## 5 1.000000 0.821869 0.000000 0.178131 4
##
## ---
## threshold f1 f2 f0point5 accuracy precision recall
## 395 0.015793 0.730200 0.871235 0.628464 0.578046 0.575051 1.000000
## 396 0.013890 0.729260 0.870700 0.627351 0.576032 0.573887 1.000000
## 397 0.011867 0.728792 0.870433 0.626796 0.575025 0.573306 1.000000
## 398 0.010070 0.728324 0.870166 0.626243 0.574018 0.572727 1.000000
## 399 0.007493 0.727856 0.869899 0.625690 0.573011 0.572149 1.000000
## 400 0.004727 0.726923 0.869365 0.624587 0.570997 0.570997 1.000000
## specificity absolute_mcc min_per_class_accuracy mean_per_class_accuracy tns
## 395 0.016432 0.097207 0.016432 0.508216 7
## 396 0.011737 0.082072 0.011737 0.505869 5
## 397 0.009390 0.073370 0.009390 0.504695 4
## 398 0.007042 0.063508 0.007042 0.503521 3
## 399 0.004695 0.051828 0.004695 0.502347 2
## 400 0.000000 0.000000 0.000000 0.500000 0
## fns fps tps tnr fnr fpr tpr idx
## 395 0 419 567 0.016432 0.000000 0.983568 1.000000 394
## 396 0 421 567 0.011737 0.000000 0.988263 1.000000 395
## 397 0 422 567 0.009390 0.000000 0.990610 1.000000 396
## 398 0 423 567 0.007042 0.000000 0.992958 1.000000 397
## 399 0 424 567 0.004695 0.000000 0.995305 1.000000 398
## 400 0 426 567 0.000000 0.000000 1.000000 1.000000 399