我不知道为什么,对于那个简单的例子,我无法让Lime工作。这是显示问题的独立代码:
data(Theoph)
library(caret)
outcomeName <- 'conc'
predictors<-names(Theoph)[!names(Theoph) %in% outcomeName]
model_gbm <- train(Theoph[,predictors],Theoph[,outcomeName],method='gbm')
library(lime)
library(tidyr)
explainer_caret <- lime(x = as.data.frame(Theoph[, predictors]), model = model_gbm, n_bins = 10)
explanation_caret <- explain(
x = as.data.frame(Theoph[c(1), predictors]),
explainer = explainer_caret,
dist_fun = "gower",
kernel_width = NULL,
n_features = 2,
feature_select = "auto"
)
glm.fit(x = x_fit,y = y,weights = weights,family = gaussian())错误: NA / NaN / Inf in'y'
有任何想法/建议吗?