我正在使用“插入符号”程序包对过度分散的计数数据集在负二项式回归模型（glm.nb）上执行K折交叉验证。我认为由于警告，我使用的方法不正确。
是否有一种方法可以对使用glm.nb函数创建的模型执行交叉验证？评论受到高度赞赏。
代码
data_ctrl <- trainControl(method = "cv", number = 5)
# Run the negtaive binomial regression model
model_caret <- train(Count~Soil.Sal+Temp+Soil.ph, data = CLAB,trControl=data_ctrl, method="lm")
model_caret
model_caret$finalModel
输出
Negative Binomial Generalized Linear Model
102 samples
3 predictor
No pre-processing
Resampling: Cross-Validated (5 fold)
Summary of sample sizes: 81, 82, 82, 82, 81
Resampling results across tuning parameters:
link RMSE Rsquared MAE
identity NaN NaN NaN
log 7.680947 0.3695579 5.186753
sqrt 6.727719 0.3854733 4.751341
RMSE was used to select the optimal model using the smallest value.
The final value used for the model was link = sqrt
Warnings (12)
Warning messages:
1: In log(y/mu) : NaNs produced
2: model fit failed for Fold1: link=identity Error : no valid set of coefficients has been found: please supply starting values