suppose that we take a data set and divide it into two equal parts at random, namely training and testing sets. we try out two different classification predictive models: model 1 and model 2. first, you use model 1 and get an error rate of 35% on the training data and 40% on the testing data. second, you use model 2 and get an error rate of 5% on the training data and 40% on the testing data.