Unit 4: Unsupervised Learning - Practice Quiz

1 What is the main goal of clustering?

2 In K-Means clustering, what does represent?

3 What is a centroid in K-Means clustering?

4 How does K-Means usually assign a data point to a cluster?

5 Which diagram commonly displays the result of hierarchical clustering?

6 What does agglomerative hierarchical clustering do first?

7 What type of clustering algorithm is DBSCAN?

8 How does DBSCAN commonly treat points in sparse regions?

9 What is the main purpose of dimensionality reduction?

10 What does PCA stand for?

11 What does the first principal component capture?

12 Principal components produced by PCA are typically:

13 How can reducing data to two principal components help a student?

14 What is the main purpose of association rule mining?

15 In the rule , what does the arrow indicate?

16 What does support measure in association rule mining?

17 What is the Apriori algorithm mainly used to find?

18 According to the Apriori principle, if an itemset is frequent, its subsets must be:

19 How can clustering help a robot interpret sensor data?

20 Which task is a common unsupervised learning application for sensor data?

21 A data analyst applies K-Means clustering with to customer data. After assigning points to clusters, which step occurs next during a standard iteration?

22 Why is feature scaling usually important before applying K-Means to variables such as income and age?

23 A hierarchical clustering dendrogram shows that two groups merge only at a very large distance. What does this most likely indicate?

24 Which situation is most suitable for DBSCAN rather than K-Means?

25 In DBSCAN, a point has at least the required number of neighbors within the radius . How is this point generally classified?

26 A dataset contains 50 correlated sensor measurements. What is a primary reason to apply PCA before visualization?

27 What does the first principal component represent in PCA?

28 Why are features commonly standardized before PCA when they use different units?

29 A PCA model retains three components explaining 72%, 18%, and 5% of the total variance. What percentage of variance is discarded if only these components are retained?

30 Which trade-off is most directly associated with reducing 100 features to 10 principal components?

31 In association rule mining, a rule has support 0.30 and confidence 0.75. What does the confidence value indicate?

32 A rule has support 0.20 and confidence 0.80. If the support of its antecedent is 0.25, which calculation gives the support of the rule's consequent relationship?

33 What is the main purpose of the Apriori property in frequent itemset mining?

34 If the itemset is below the minimum support threshold, what should Apriori do with any larger candidate containing both and ?

35 A store discovers the rule {bread, butter} {jam}. Which action best uses this pattern?

36 A robot receives unlabeled readings from temperature, vibration, and motor-current sensors. How can clustering support predictive maintenance?

37 A mobile robot uses unsupervised learning to group range-sensor readings. What is a reasonable use of the resulting groups?

38 A robot's accelerometer data is reduced from six dimensions to two using PCA. What is the most likely benefit?

39 A robot records sensor readings during normal operation but has no fault labels. Which unsupervised method is appropriate for detecting possible anomalies?

40 Several robot sensors measure related aspects of the same movement, producing highly correlated features. Why might PCA be useful before clustering?

41 A dataset contains two features: annual income ranging from to and age ranging from to . A k-means model using Euclidean distance is trained without preprocessing. Which outcome is most likely?

42 In k-means clustering, two clusters become empty after an assignment step. Which response preserves the basic optimization objective most appropriately?

43 A k-means algorithm repeatedly converges to different solutions when its initial centroids change. Which interpretation is most accurate?

44 In agglomerative hierarchical clustering, four points form two compact groups that are far apart. Which linkage is most likely to merge the groups prematurely because of a single pair of close points?

45 For DBSCAN, a point is within the neighborhood radius of two core points belonging to different clusters, but it is not itself a core point. What can happen?

46 The centered covariance matrix of a two-dimensional dataset is . If one principal component is retained, which proportion of total variance is preserved?

47 Two features have covariance matrix . Which direction is the first principal component most likely to follow?

48 A PCA model is fit using all available data before splitting observations into training and test sets. Why can this produce an overly optimistic downstream evaluation?

49 A dataset contains one feature measured in millimeters and another measured in kilograms. Their raw variances differ by a factor of , but both physical quantities are considered equally important. Which preprocessing is most defensible before PCA?

50 A two-dimensional dataset lies close to a curved one-dimensional manifold. PCA retains one component, but the reconstruction error remains high. Which conclusion is most justified?

51 In a transaction database, of transactions contain bread, contain butter, and contain both. What are the confidence and lift of the rule ?

52 Apriori uses the downward-closure property. If the itemset is below the minimum support threshold, which statement must be true?

53 Suppose the frequent 2-itemsets are , , , , and , but is infrequent. Which candidate 3-itemset is eliminated by Apriori before database scanning?

54 A rule has high confidence because its consequent occurs in of all transactions. Which metric is most useful for checking whether the rule provides meaningful association beyond the consequent's baseline frequency?

55 An itemset has support , and every proper subset with the same items removed has support greater than . Which description best fits this itemset under standard terminology?

56 A mobile robot clusters 2D lidar endpoints using DBSCAN. Increasing the radius substantially causes nearby walls and separate obstacles to merge. Which change most directly reduces this failure while preserving density-based clustering?

57 A robot uses PCA to compress a 100-dimensional sensor vector. A rare collision signal lies in a low-variance direction and disappears after retaining only the first five components. What is the main risk?

58 A robot clusters temperature and vibration readings collected over several months. A new operating regime gradually changes both distributions, causing old clusters to absorb new observations poorly. Which issue is most directly involved?

59 A robot combines range, camera color, and inertial features in k-means. The inertial variables have much larger numerical variance than the others. Which remedy is most appropriate before clustering?

60 A robot uses unsupervised clustering to group spatial lidar points before tracking objects. Two objects temporarily overlap from the sensor's viewpoint and form one dense cluster. What limitation does this illustrate?