Unsupervised learning algorithms are used for when you have the input features, X, and the labels, y.

Question 1 options:

	True
	False

Question 2(1 point)

Listen

Unsupervised learning algorithms are used for when you have the input features, X, but do not have the labels, y.

Question 2 options:

	True
	False

Question 3(1 point)

Listen

Clustering is a(n) _____ learning task.

Question 3 options:

	supervised
	unsupervised
	semi-supervised
	reinforcement

Question 4(1 point)

Listen

If y_pred and y are numpy arrays of equal size, then what is returned from the following function call?

numpy.sum(y_pred==y)

Question 4 options:

	a numpy array of equal size with numeric values
	a Boolean-valued numpy array of equal size
	the integer value of 1 or 0
	some positive integer value

Question 5(1 point)

Listen

The "K" in the K-Means clustering algorithm must be specified by the user prior to fitting the model.

Question 5 options:

	True
	False

Question 6(1 point)

Listen

The "K" in the K-Means clustering algorithm refers to the number of what?

Question 6 options:

	instances in the data set
	K-Means models that are actually fit by the Scikit-Learn KMeans class implementation
	clusters
	iterations used for convergence

Question 7(1 point)

Listen

The "label" assigned to an instance by the K-Means algorithm is semantically identical to the "label" of an instance classification problem.

Question 7 options:

	True
	False

Question 8(1 point)

Listen

Suppose you are trying to run a clustering algorithm on data that exhibits 10 nearly spherical clusters with some noise in the system. The diameter of these clusters, relative to the diameters of the other clusters, bears little to no influence on the accuracy of the clustering algorithm.

Question 8 options:

	True
	False

Question 9(1 point)

Listen

The transform() method of the KMeans class in Scikit-Learn does which of the following?

Question 9 options:

	measures the distance from each observation to every cluster centroid
	measured the squared distance from each observation to every cluster centroid
	transforms the training data to a sequence of cluster labels
	transforms the training data to a sequence of cluster scores

Question 10(1 point)

Listen

The K-Means clustering algorithm is sensitive to the choice of randomly initialized centroid locations, meaning you may get very different models based on where you initialize the centroid locations.

Question 10 options:

	True
	False

Question 11(1 point)

Listen

If you happen to know approximately where the centroids should be, then you can initialize them manually by using the "n_init" hyperparameter.

Question 11 options:

	True
	False

Question 12(1 point)

Listen

If you happen to know approximately where the centroids should be, then you can initialize them manually by using the "init" hyperparameter.

Question 12 options:

	True
	False

Question 13(1 point)

Listen

A K-Means model's inertia is the mean squared distance between each instance and every other instance.

Question 13 options:

	True
	False

Question 14(1 point)

Listen

A K-Means model's inertia is the mean squared distance between each instance and its closest centroid.

Question 14 options:

	True
	False

Question 15(1 point)

Listen

The KMeans class runs the algorithm a number of times specified by the "n_init" hyperparameter and keeps the one with the lowest silhouette score.

Question 15 options:

	True
	False

Question 16(1 point)

Listen

The KMeans class runs the algorithm a number of times specified by the "n_init" hyperparameter and keeps the one with the lowest inertia.

Question 16 options:

	True
	False

Question 17(1 point)

Listen

In an attempt to minimize the likelihood of converging to a suboptimal solution, the K-Means++ algorithm employs a random initialization algorithm that tends to select centroids that are distant from one another.

Question 17 options:

	True
	False

Question 18(1 point)

Listen

In an attempt to minimize the likelihood of converging to a suboptimal solution, the K-Means++ algorithm employs a random initialization algorithm that tends to select centroids that are very near to each other.

Question 18 options:

	True
	False

Question 19(1 point)

Listen

The K-Means algorithm, along with a few (but not all) of its variants, are able to determine the optimal number of clusters to be used, K.

Question 19 options:

	True
	False

Question 20(1 point)

Listen

The inertia performance metric can be used to determine the optimal number of clusters for a given dataset.

Question 20 options:

	True
	False

Question 21(1 point)

Listen

The silhouette score is the mean silhouette coefficient over all instances, and is a good measure for determining the optimal number of clusters to use.

Question 21 options:

	True
	False

Question 22(1 point)

Listen

Scaling of input features is not required for the K-Means algorithm

Question 22 options:

	True
	False

Question 23(1 point)

Listen

Active learning is when a learning algorithm incorporates the expertise of a human in the learning process.

Question 23 options:

	True
	False

Question 24(1 point)

Listen

The DBSCAN algorithm defines clusters as continuous regions of high density.

Question 24 options:

	True
	False

Question 25(1 point)

Listen

What is a core instance of the DBSCAN clustering algorithm?

Question 25 options:

	an observation with some required number of other instances within its epsilon neighborhood
	any instance that is not considered to be an anomaly by the algorithm
	any observation within the epsilon neighborhood of another instance
	both A and B

Question 26(1 point)

Listen

When the DBSCAN algorithm assigns an index of "-1" to an instance, it means that instance belongs to the cluster with the label of "-1".

Question 26 options:

	True
	False

Question 27(1 point)

Listen

Question 27 options:

In the case of

clustering, each instance is assigned to a single cluster. In the case of

clustering, each instance is given a score per cluster.

Question 28(1 point)

Listen

Question 28 options:

	True
	False

Question 29(1 point)

Listen

Question 29 options:

The

K-Means algorithm improves the traditional K-Means algorithm by avoiding unnecessary distance calculations.

Question 30(1 point)

Listen

Question 30 options:

Suppose there was an image of fruits and vegetables covering a table. If you wanted to segment each individual grape separately in the entire image, then you would use

segmentation.

Question 31(1 point)

Listen

Question 31 options:

Suppose there was an image of fruits and vegetables covering a table. If you wanted to segment all of the grapes in the entire image at once, then you would use

segmentation.