Instructions. If any of the instructions are disobeyed, candidates shall expose themselves to disqualification from future examinations.

• General: All questions must be answered on BC Connect. Before clicking submit at the end of the examination, make sure all items you completed shows as completed. 
• General: No instructions or directives of the invigilator shall be disregarded; if any of the instructions are disobeyed, candidates shall expose themselves to disqualification from future examinations. 
• General: Each candidate is expected to maintain academic integrity. Any form of plagiarism or cheating that is detected can lead to disciplinary measures being taken.  
• General: No explanation of the examination questions may be asked or shall be given.
• General: Late submissions will not be accepted without proof.
• General: Candidates shall not communicate or attempt to communicate with anyone and candidates shall not conduct themselves in an improper or unseemly manner. 
• General: Should any work need to be uploaded, ensure naming conventions are followed as instructed.
• Technical: Any technical issue related to the examination platform being used needs to be reported to the invigilator immediately. 
• Mathematical: In the event of a calculation error on a question, marking will not continue due to method marking not being applied. Note, for all non-integer values, round them off to what is specified in the question. Finite values should be left as they are. However if no instruction is given and the values are not exact, round to 4 significant figures.
• Mathematical: Calculators (no online calculators)and Excel can be used.
• Mathematical: For all questions where you need to fill in gaps, only use lowercase letters for any words or variables unless something else was specified. You will see that only dot is allowed for decimals and if necessary only "/" can be used for fractions, but read the instructions carefully.
• Online: Online students must be in the correct virtual room with their webcams and visible during the entire examination. Online students need to share their entire screen. Online students must ensure they do not encounter any interruptions. When you are done, you need to notify the invigilator before leaving the meeting. No headphones or earphones are allowed.
• Practical: All answers must be submitted before the conclusion of the examination. If the examination has a timer, make sure to refresh the page before checking how much time you have left if you started at the time that the invigilator started to time the examination.

• Important Instruction for Students Using Excel

  If you choose to complete the calculations using Excel, you must only use the formulas exactly as they are provided on the lecture slides.

  Please note that Excel contains many built-in formula functions that may produce slightly different answers due to rounding or calculation methods. For the purpose of this assessment, we will only accept answers that are obtained using the formulas provided on the slides.

  Markers will not apply a range of acceptable answers. Therefore, if you use Excel functions or formulas that differ from those shown on the slides and your answer varies even slightly from the expected result, no marks will be awarded for that question.

Make sure you pay extra attention to the instructions given to each question in this paper.

Write then submit before the indicated time expires or before the time is up from when the invigilator started to time the session.

Logistic Regression Probability Model

Consider a logistic regression model. What is the general expression for the probability P(Y = 1 | X = x, θ), where:

• are the features (with X₀ = 1 for the bias term),
• is a binary outcome (0 or 1),
• is the parameter vector.

Choose the correct expression:

In which of the following situations would it be appropriate to use logistic regression?

You are designing a Naïve Bayes classifier to predict whether a customer will buy a computer (Buy_Computer = Yes or No) based on a feature tuple

Bayes’ theorem gives the posterior probability:

When classifying a new customer, why is the division by P(X) often omitted when comparing P(Buy_Computer = Yes | X) and P(Buy_Computer = No | X)?

Select one:

Support vector machines, like logistic regression models, give a probability distribution over the possible labels given an input example.

In SVMs, what is the role of the kernel function ?

The K-Means objective function minimizes:

In logistic regression, the decision boundary is:

In a classification tree using the CART algorithm, which criterion is typically minimized at each split?

Which of the following statements about Information Gain is correct?