Description
Problem 1 In this problem you will compare the performance of a variety of classifiers that you have learned
about so far. The data is in the file magic04.data and the column names are in the file magic04.names. The
last column is a categorical response with values g or h, and the rest of the columns are numerical features.
You can read more about the dataset here.
(a) Load the data (can use the pandas function read table with the arguments sep=’,’ and header=None).
Split the data into a training and test set. Scale and center the columns using the mean and standard
deviation of each column from the training set (make sure you use the same scaling on the test set
that is used on the training set).
(b) Learn the following models to classify the training data:
• Logistic Regression: Can import LogisticRegression from sklearn.linear model.
• LDA: Can import LinearDiscriminantAnalysis from sklearn.discriminant analysis.
• KNN Classifier: Need to choose the number of neighbors k.
• Linear SVM: Need to choose the margin penalty C as a hyperparameter.
• Gaussian (Radial) SVM: Need to choose the margin penalty C and the radius width γ.
To tune hyperparameters for each model, you can either use cross-validation or hand-tune by examining
the model performance for reasonable values of the hyper-parameters.
(c) Apply your models to the test set. Report the accuracy, visualize an ROC curve, and report the AUC
for each model. For Logistic Regression, report the most meaningful predictors.
1
