Question 1 10 pts Question 1: True/False The training set is used to estimate the error rate of the trained classier. 0 True 0 False Question 2 10 pts Question 2: Fill in the blanks If a model has _____ training set error and _____ test set error, the model overts. 0 low, high 0 low, low 0 high, low 0 high, high Question 3 10 pts Question3: Suppose we are given a dataset of 500 observations, 500 models are trained with 499 observations as part of the training set. and the remaining 1 observation for testing. The error rate is averaged out. This validation technique is known as O Holdout method 0 LOOCV O Bootstrap O K-fold cross validation Question 4 10 pts Question 4: Which of the following statement is TRUE with respect to the holdout and LOOCV method? 0 If you repeat the holdout method many times, you will always get the same estimate for test error. 0 None of the above. 0 The holdout method splits the dataset into two equal-sized groups. 0 LOOCV is computationally less expensive than the holdout method. question 3 10 pts Question 5: Suppose we want to develop a machine learning algorithm that classies whether a student will pass or fail the exam based on several factors (e.g., number of hours spent per week, past homework scores), then which of the following metrics would you choose to evaluate the classier? O MSE 0 Both of them 0 Accuracy Question 6 10 pts Question 6: Which of the following statements is/are TRUE for Kfold cross validation? 1) A larger K means that each model is trained on a smaller training set and tested on a larger test set. 2) When K = N, Kfold cross validation is equivalent to LOOCV where N is the number of observations. 3) For any K, repeated application of Kfold cross validation will always produce the same error. 4) Kfold cross validation is a resampling method WITH replacement. 0 Only 2) O 3) and 4) O 1), 2), 3) and 4) O 1) and 2) Question 7: Suppose we have a dataset of 500 observations, and we want to use a 10-fold cross validation for model selection, then how many models do you need to train and how large is your training set? 0 10 models, 500 observations 0 1 model, 450 observations 0 1 model, 50 observations 0 10 models, 450 observations Consider the following dataset: X = {26, 30, 17, 29, 25, 23} - The rst bootstrap yields the dataset {30, 26, 17, 26, 25, 23} - The second bootstrap yields the dataset {26, 26, 26, 29, 17, 26} - The third bootstrap yields the dataset {17, 26, 29, 29, 17,23} Question 8 10 p Question 8: Compute the bias of the sample mean. (Please keep 2 decimal places.) Question 9 10 pts Question 9: Compute the variance of the sample mean. (Please keep 2 decimal places.) Question 10 10 pts Question 10: Which of the following validation techniques may not be suitable for large datasets with hundreds of thousands of samples? 0 LOOCV O K-fold cross validation O Holdout method O Bootstrap