15 01 21

Page 1

15/1/2021 2021.01.15 Machine Learning BIO https://forms.office.com/Pages/ResponsePage.aspx?id=K3EXCvNtXUKAjjCd8ope66_XbeATVphMnuFlOS05Oj1UMzc3WENFWVVITjUwTTdIMjkyMjR… 1/6 2021.01.15 Machine Learning BIO  Ciao Davide, quando invierai il modulo, il proprietario potrà vedere il tuo nome e l'indirizzo di posta elettronica. 1 In the following box you can add any comment related to the test. (withdraw, clarification, etc) Inserisci la risposta 2 Which ones among the following answers is correct? Independence of the residuals can be verified by the K olmogorov-Smirnov test. In K-NN for regression the predicted target value is the average of the targets for the K nearest neighbors. In regression trees the predicted target value in a leaf is the variance of the target for the observations in the leaf itself. Least squares regression may be ill-conditioned due to multi-collinearity. Backward variable selection in least squares regression is a filtering technique. None of the other answers are correct. 15/1/2021 2021.01.15 Machine Learning BIO https://forms.office.com/Pages/ResponsePage.aspx?id=K3EXCvNtXUKAjjCd8ope66_XbeATVphMnuFlOS05Oj1UMzc3WENFWVVITjUwTTdIMjkyMjR… 2/6 3 The figure depicts the training set of a binary classification problem in a dataset with two numerical explanatory features: blue and red points correspond to classes 0 and 1 respectively. Which ones among the following models CAN NO T reach a perfect classification in this example? http://www.dim.uchile.cl/~mausoto/images/ml2/data_1.png None (all models can achieve a perfect classification in the example). Classification T ree of depth 3. K-NN. Logistic R egression SVM with a linear kernel. Multilayer Neural Network. 15/1/2021 2021.01.15 Machine Learning BIO…