This report aims to present the capabilities of the package DALEXtra.

The document is a part of the paper “Landscape of R packages for eXplainable Machine Learning”, S. Maksymiuk, A. Gosiewska, and P. Biecek. (https://arxiv.org/abs/2009.13248). It contains a real life use-case with a hand of titanic_imputed data set described in Section Example gallery for XAI packages of the article.

We did our best to show the entire range of the implemented explanations. Please note that the examples may be incomplete. If you think something is missing, feel free to make a pull request at the GitHub repository MI2DataLab/XAI-tools.

The list of use-cases for all packages included in the article is here.

Load titanic_imputed data set.

data(titanic_imputed, package = "DALEX")

head(titanic_imputed)

##   gender age class    embarked  fare sibsp parch survived
## 1   male  42   3rd Southampton  7.11     0     0        0
## 2   male  13   3rd Southampton 20.05     0     2        0
## 3   male  16   3rd Southampton 20.05     1     1        0
## 4 female  39   3rd Southampton 20.05     1     1        1
## 5 female  16   3rd Southampton  7.13     0     0        1
## 6   male  25   3rd Southampton  7.13     0     0        1

Fit a forest type model and logistic regression to the titanic imputed data.

ranger_model <- ranger::ranger(survived~., data = titanic_imputed, classification = TRUE, probability = TRUE)
gbm_model <- gbm::gbm(survived~., data = titanic_imputed, distribution = "bernoulli")

library(DALEXtra)

Model Diagnostics

Funnel Plot

explainer_ranger <- explain(ranger_model, data = titanic_imputed, y = titanic_imputed$survived, label = "Ranger Model")

## Preparation of a new explainer is initiated
##   -> model label       :  Ranger Model 
##   -> data              :  2207  rows  8  cols 
##   -> target variable   :  2207  values 
##   -> predict function  :  yhat.ranger  will be used ( [33m default [39m )
##   -> predicted values  :  numerical, min =  0.01559981 , mean =  0.322142 , max =  0.9888994  
##   -> model_info        :  package ranger , ver. 0.12.1 , task classification ( [33m default [39m ) 
##   -> residual function :  difference between y and yhat ( [33m default [39m )
##   -> residuals         :  numerical, min =  -0.7815942 , mean =  1.47828e-05 , max =  0.8801861  
##  [32m A new explainer has been created! [39m

explainer_gbm <- explain(gbm_model, data = titanic_imputed, y = titanic_imputed$survived, label = "GBM Model")

## Preparation of a new explainer is initiated
##   -> model label       :  GBM Model 
##   -> data              :  2207  rows  8  cols 
##   -> target variable   :  2207  values 
##   -> predict function  :  yhat.gbm  will be used ( [33m default [39m )
##   -> predicted values  :  numerical, min =  0.05005481 , mean =  0.3223963 , max =  0.9660827  
##   -> model_info        :  package gbm , ver. 2.1.8 , task classification ( [33m default [39m ) 
##   -> residual function :  difference between y and yhat ( [33m default [39m )
##   -> residuals         :  numerical, min =  -0.9660827 , mean =  -0.0002395298 , max =  0.9470355  
##  [32m A new explainer has been created! [39m

fm <- funnel_measure(explainer_gbm, explainer_ranger, show_info = FALSE)
plot(fm)

## $`challanger_Ranger Model`

Overall comaprison

explainer_ranger <- explain(ranger_model, data = titanic_imputed, y = titanic_imputed$survived, label = "Ranger Model", verbose = FALSE)
explainer_gbm <- explain(gbm_model, data = titanic_imputed, y = titanic_imputed$survived, label = "GBM Model", verbose = FALSE)

oc <- overall_comparison(explainer_gbm, explainer_ranger, type = "classification")
plot(oc)

## $radar_plot

## 
## $accordance_plot

Training test comparison

ind <- sample(1:nrow(titanic_imputed), 0.7*nrow(titanic_imputed))
train <- titanic_imputed[ind,]
test <- titanic_imputed[-ind,]

ranger_model <- ranger::ranger(survived~., data = train, classification = TRUE, probability = TRUE)
gbm_model <- gbm::gbm(survived~., data = train, distribution = "bernoulli")

explainer_ranger <- explain(ranger_model, data = test, y = test$survived, label = "Ranger Model", verbose = FALSE)
explainer_gbm <- explain(gbm_model, data = test, y = test$survived, label = "GBM Model", verbose = FALSE)

tt <- training_test_comparison(explainer_gbm, explainer_ranger, training_data = train, training_y = train$survived)
plot(tt)

Report

Compiled report can be seen under this link: https://mi2datalab.github.io/IML-tools/DALEXtra_champion_challenger.html.

explainer_ranger <- explain(ranger_model, data = titanic_imputed, y = titanic_imputed$survived, label = "Ranger Model")
explainer_gbm <- explain(gbm_model, data = titanic_imputed, y = titanic_imputed$survived, label = "GBM Model")

fm <- funnel_measure(explainer_gbm, explainer_ranger, show_info = FALSE)

explainer_ranger <- explain(ranger_model, data = titanic_imputed, y = titanic_imputed$survived, label = "Ranger Model", verbose = FALSE)
explainer_gbm <- explain(gbm_model, data = titanic_imputed, y = titanic_imputed$survived, label = "GBM Model", verbose = FALSE)

oc <- overall_comparison(explainer_gbm, explainer_ranger, type = "classification")

ind <- sample(1:nrow(titanic_imputed), 0.7*nrow(titanic_imputed))
train <- titanic_imputed[ind,]
test <- titanic_imputed[-ind,]

ranger_model <- ranger::ranger(survived~., data = train, classification = TRUE, probability = TRUE)
gbm_model <- gbm::gbm(survived~., data = train, distribution = "bernoulli")

explainer_ranger <- explain(ranger_model, data = test, y = test$survived, label = "Ranger Model", verbose = FALSE)
explainer_gbm <- explain(gbm_model, data = test, y = test$survived, label = "GBM Model", verbose = FALSE)

tt <- training_test_comparison(explainer_gbm, explainer_ranger, training_data = train, training_y = train$survived)

champion_challenger(list(fm, oc, tt), title = "DALEXtra: Champion - Challenger analysis", author = "Szymon Maksymiuk", model_performance_table = TRUE, output_name = "DALEXtra_champion_challenger")

Session info

sessionInfo()

## R version 3.6.1 (2019-07-05)
## Platform: x86_64-w64-mingw32/x64 (64-bit)
## Running under: Windows 10 x64 (build 18363)
## 
## Matrix products: default
## 
## locale:
## [1] LC_COLLATE=Polish_Poland.1250  LC_CTYPE=Polish_Poland.1250   
## [3] LC_MONETARY=Polish_Poland.1250 LC_NUMERIC=C                  
## [5] LC_TIME=Polish_Poland.1250    
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
## [1] DALEXtra_2.0 DALEX_2.0.1 
## 
## loaded via a namespace (and not attached):
##  [1] Rcpp_1.0.4.6       compiler_3.6.1     pillar_1.4.4       tools_3.6.1       
##  [5] digest_0.6.25      jsonlite_1.6.1     evaluate_0.14      lifecycle_0.2.0   
##  [9] tibble_3.0.1       gtable_0.3.0       lattice_0.20-40    pkgconfig_2.0.3   
## [13] rlang_0.4.6        Matrix_1.2-18      ggrepel_0.8.2      yaml_2.2.1        
## [17] gbm_2.1.8          xfun_0.12          gridExtra_2.3      ranger_0.12.1     
## [21] stringr_1.4.0      dplyr_0.8.5        knitr_1.28         vctrs_0.3.1       
## [25] grid_3.6.1         tidyselect_1.0.0   reticulate_1.14    glue_1.4.1        
## [29] R6_2.4.1           survival_3.1-11    rmarkdown_2.1      hnp_1.2-6         
## [33] auditor_1.2.1.0000 farver_2.0.3       ggplot2_3.3.2      purrr_0.3.4       
## [37] magrittr_1.5       MASS_7.3-51.6      scales_1.1.1       htmltools_0.4.0   
## [41] ellipsis_0.3.1     splines_3.6.1      assertthat_0.2.1   colorspace_1.4-1  
## [45] labeling_0.3       stringi_1.4.6      munsell_0.5.0      crayon_1.3.4

The DALEXtra R package