[R] Statistical tests on model trained on different datasets

[Neha gupta]

Hi

If I have one dataset , then we can perform statistical tests as:

d=read.csv("dataset1)

svm <- train(bug ~ ., data = tr,
            --
---
getTrainPerf(svm)

rf <- train(bug ~ ., data = tr,
            --
---
getTrainPerf(rf)

wilcox.test(svm$resamples$RMSE, rf$resamplesRMSE)

But if we have another model trained with another dataset

dd=  read.csv("dataset2")

gbm<- train(bug ~ ., data = tr,
            --
---
getTrainPerf(gbm)

How can I perform wilcox test between rf and gbm, which are trained on
different datasets.

Thanks

	[[alternative HTML version deleted]]