We are facing a prediction problem that follows the standard conditions. y /(x) +c. independently sampled observations, However, we are in a special case, the variance of the noise Var(c) is zero. We are given the training errors for a set of candidate models Each model has a different training error, one of them gets a perfect fit, training error of zero. Select ALL that are TRUE. The model with training error of zero has higher variance than the others. (in the Bias-Variance decomposition of Expected Prediction Error) () The model with training error of zero has greater evidence of resulting in better Prediction Error The models with training error greater than zero have blas greater than zero (in the Bias-Variance decompostion of Expected Prediction Error)