1. List three Algorithms or Models that can be used for Classification, and critically evaluate their merits. 2. Since many real-world relationships are not linear, should you always consider non-linear models such as high-degree polynomial functions instead of the basic linear regression models? Why? Please form your arguments based on theoretical considerations. However, empirical evidence is also acceptable.

3. What is the 'Curse of Dimensionality'? What problems could high dimensional data cause for the K-NN and the Decision Tree algorithms? In addition, what are the main motivations for reducing a dataset's dimensionality?

4. Prof. Ghahramani (Cambridge University) once said that one of his long-term research projects was to build an "automated statistician". What do you think that means? In addition, what do you think would go into building one and what are the major challenges?

5. How do Fitting graphs and Learning curves address the Overfitting problem? In addition, suppose you are using a multiple high-degree polynomial regression model to learn the data, and the Fitting graph shows that the testing data error is much higher than the training data error, please propose four possible solutions to address this.

6. Why would data-miners sometimes prefer to use a regularised model instead of a plain linear regression model? In addition, among the three regularisation methods you learnt from this module (RR, Lasso, E-N), discuss their data and model specification suitability