The EIX R package
08-10-2020
This report aims to present the capabilities of the package EIX.
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 1library(xgboost)
library(EIX)Fit a xgboost to the titanic imputed data.
library("Matrix")
X <- sparse.model.matrix(survived ~ . - 1, data = titanic_imputed)
xgb_model <- xgboost(X, titanic_imputed$survived, nrounds = 10, params = list(objective = "binary:logistic"), prediction = TRUE)## [1] train-error:0.168555
## [2] train-error:0.168101
## [3] train-error:0.168101
## [4] train-error:0.167195
## [5] train-error:0.164024
## [6] train-error:0.160852
## [7] train-error:0.157227
## [8] train-error:0.157680
## [9] train-error:0.153149
## [10] train-error:0.149977Model parts
Interaction measure
interactions <- interactions(xgb_model, X, option = "interactions")
plot(interactions)
Variable Importance
importance <- importance(xgb_model, X, option = "both")
plot(importance)
Predict parts
Interaction based explanation of single observation
new_observation <- X[27,]
wf <- waterfall(xgb_model, new_observation, titanic_imputed[27,], option = "interactions")
plot(wf)
Model diagnostics
lolli <- lollipop(xgb_model, X)
plot(lolli, threshold=0.02)
Session info
sessionInfo()## R version 4.0.2 (2020-06-22)
## Platform: x86_64-w64-mingw32/x64 (64-bit)
## Running under: Windows 10 x64 (build 17763)
##
## 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] Matrix_1.2-18 EIX_1.1 xgboost_0.90.0.2
##
## loaded via a namespace (and not attached):
## [1] tidyselect_1.1.0 sjlabelled_1.1.6 xfun_0.12 mycor_0.1.1
## [5] purrr_0.3.3 reshape2_1.4.3 splines_4.0.2 lattice_0.20-40
## [9] colorspace_1.4-1 vctrs_0.3.2 generics_0.0.2 htmltools_0.4.0
## [13] yaml_2.2.1 mgcv_1.8-31 rlang_0.4.7 pillar_1.4.4
## [17] glue_1.4.1 gdtools_0.2.2 RColorBrewer_1.1-2 uuid_0.1-4
## [21] lifecycle_0.2.0 plyr_1.8.6 stringr_1.4.0 sjmisc_2.8.5
## [25] munsell_0.5.0 gtable_0.3.0 htmlwidgets_1.5.1 evaluate_0.14
## [29] labeling_0.3 DALEX_1.3.1.1 knitr_1.28 iBreakDown_1.2.0
## [33] ggiraph_0.7.8 Rcpp_1.0.4.6 scales_1.1.1 ppcor_1.1
## [37] farver_2.0.3 systemfonts_0.2.3 ggplot2_3.3.2 digest_0.6.25
## [41] stringi_1.4.6 ggrepel_0.8.2 insight_0.9.0 dplyr_1.0.1
## [45] grid_4.0.2 tools_4.0.2 magrittr_1.5 tibble_3.0.1
## [49] crayon_1.3.4 tidyr_1.0.2 pkgconfig_2.0.3 ellipsis_0.3.1
## [53] MASS_7.3-51.6 data.table_1.12.8 rmarkdown_2.1 R6_2.4.1
## [57] ggiraphExtra_0.2.9 nlme_3.1-140 compiler_4.0.2