For boston housing dataset, conduct 2-fold CV by cv.glm to calculate the mean square error of full model, and a reduced model with indus and rm only. In terms of the CV criterion, which of the two models do you prefer?
cv.glm