Machine Learning Quiz 3

The logistic function is a sigmoid function that assumes values from __ to __.

0, 1

You are presented with a training dataset of 100 instances with two class labels: A and B. The class distribution between labels A and B is 80 to 20, respectively. Prior to building a model, you decide to balance the training data. Using the SMOTE() function, you set perc.over = 50 and perc.under = 300. After balancing the training set, you will now have ____________ instances of class label A and ____________ instances of class label B.

30,30

A common type of distance measure used in kNN is the ____________ distance.

Euclidean

Given a set of candidate models from the same data, the model with the highest AIC is usually the "preferred" model.

False

In binomial logistic regression, the best cut-off point is always at 0.5

False

Regression establishes causation between the independent variables and the dependent variable.

False

In order to choose the next feature to split on, a decision tree learner calculates the Information Gain for features A, B, C and D as 0.022, 0.609. 0.841 and 0.145 respectively. Which feature will it next choose to split on?

Feature C (0.841)

Rather than the sum of squares used in linear regression, in logistic regression, the coefficients are estimated using a technique called ____________.

Maximum Likelihood Estimation.

The k in kNN refers to ____________.

The number of labeled observations to compare with the unlabeled observation.

Decision tree learners typically output the resulting tree structure in human-readable format. This makes them well suited for applications that require transparency for legal reasons or for knowledge transfer.

True

Entropy is highest when the split is 50-50. However, as one class dominates the other, entropy reduces towards zero.

True

Features with a large number of distinct values will have lower intrinsic value than features with a small number of distinct values.

True

Before we use kNN, what can we do if we have significant variance in the range of values for our features?

We normalize the data.

You are presented with a training dataset called greek_train. It has the following features: alpha, beta, omega and zeta. To train a logistic regression model using this data, you write the following code: greek_mod <- glm(omega~. -alpha, data = greek_train, family = binomial(link = 'logit')) If you run the summary(greek_mod) command after training your model, which of the following features should you expect to get model coefficients for?

beta and zeta

Lazy learners such as k-Nearest Neighbor are also known as ____________ learners.

instance-based learners

The link function used for binomial logistic regression is called the ____________.

logit function

kNN is an example of a ____________ model.

non-parametric

A k value that is too small not only makes a model susceptible to noise and/or outliers, it can also lead to ____________.

overfitting

For decision trees, the process of remediating the size of a tree in order for it to generalize better is known as ____________.

pruning

For decision trees, entropy is a quantification of the level of ____________ within a set of class values.

randomness