PYTHON Given banknote authentication dataset. It has been split into training data banknote authentication train.csv and test data banknote authentication test.csv. File banknote feature description.csv describes the meaning of each column in the data set. a) Use k-NN with k = 2, 5 and 10 to learn the training data. Find the accuracies and F1 scores for k-NN models with different k values. Which k produces the best result? b) Use logistic regression with regularization parameter = 5 (note: must read carefully the definition the regularization parameter in sklearn). a) Explicitly write down the logistic regressions predictor (note: type down your for- mulation in the notebook) b) What is the decision boundary for this model? Plot this decision boundary along with training data. (note: you only need to type down the expression of the decision boundary) c) Evaluate accuracy, AUC, AP and F1 score for this model. Plot ROC curve and Precision-Recall curve with a full information as figures in lecture notes d) Compare this model to the previous k-NN model.(note: type down your answer in the notebook)

https://www.dropbox.com/s/fn824c882hef7yr/Screenshot%202018-03-01%2018.48.11.png?dl=0

https://www.dropbox.com/s/e7ryxhj67zg1nmd/Screenshot%202018-03-01%2018.48.31.png?dl=0

https://www.dropbox.com/s/wzh893afj1nm3rb/Screenshot%202018-03-01%2018.48.21.png?dl=0