R - model explainer

Model Explainer

#xgboost explainer
#library(devtools) 
#install_github("AppliedDataSciencePartners/xgboostExplainer")
library(xgboost)
library(xgboostExplainer)

#getting data
set.seed(123)
data(agaricus.train, package='xgboost')

X = as.matrix(agaricus.train$data)
y = agaricus.train$label

#split train test
train_idx = 1:5000
train.data = X[train_idx,]
test.data = X[-train_idx,]

#convert to matrix
xgb.train.data <- xgb.DMatrix(train.data, label = y[train_idx])
xgb.test.data <- xgb.DMatrix(test.data)

#run model
param <- list(objective = "binary:logistic")
xgb.model <- xgboost(param =param,  data = xgb.train.data, nrounds=10, early_stopping_rounds = 2)
col_names = colnames(X)
pred.train = predict(xgb.model,X)
nodes.train = predict(xgb.model,X,predleaf =TRUE)
trees = xgb.model.dt.tree(col_names, model = xgb.model)

#### The XGBoost Explainer
explainer = buildExplainer(xgb.model,xgb.train.data, type="binary", base_score = 0.5, n_first_tree= 1)

pred.breakdown = explainPredictions(xgb.model, explainer, xgb.test.data)

showWaterfall(xgb.model, explainer, xgb.test.data, test.data,  2, type = "binary")
showWaterfall(xgb.model, explainer, xgb.test.data, test.data,  8, type = "binary")



#live
#https://mi2datalab.github.io/live/
library(e1071)
library(live)
library(mlr)
library(forestmodel)
library(xgboost)
library(xgboostExplainer)
head(wine)
wine_svm <- e1071::svm(quality ~., data = wine)
similar <- sample_locally2(data = wine,
                           explained_instance = wine[10, ],
                           explained_var = "quality",
                           size = 500)
similar1 <- add_predictions2(to_explain = similar,
                             black_box_model = wine_svm)
wine_expl <- fit_explanation2(live_object = similar1,
                              white_box = "regr.lm")
plot_explanation2(wine_expl, "forest")
plot_explanation2(wine_expl, "waterfall")

#Lime
https://cran.r-project.org/web/packages/lime/vignettes/Understanding_lime.html

#XGBOOST
X <- as.matrix(wine[,1:11])
xgb.train.data <- xgb.DMatrix(X, label = wine$quality>6 )
param <- list(objective = "binary:logistic")
xgb.model <- xgboost(param =param,  data = xgb.train.data, nrounds=20, early_stopping_rounds = 2)
col_names = colnames(X)
pred.train = predict(xgb.model,X)
nodes.train = predict(xgb.model,X,predleaf =TRUE)
trees = xgb.model.dt.tree(col_names, model = xgb.model)

#### The XGBoost Explainer
explainer = buildExplainer(xgb.model,xgb.train.data, type="binary", base_score = 0.5, n_first_tree= 1)
pred.breakdown = explainPredictions(xgb.model, explainer, xgb.train.data)
showWaterfall(xgb.model, explainer, xgb.train.data, X,  100, type = "binary",  threshold = 1e-04, limits = c(NA, NA))