NEW
Font size
WorksheetsIT 311 - Final Exam
Total questions: 60
Worksheet time: 3600secs
Which of the following best describes the main purpose of cluster analysis in the context of database systems?
To divide data into groups that are meaningful, useful, or both
To encrypt data for security purposes
To delete unnecessary data from databases
To convert data into different formats
Which of the following best describes a cluster in data mining?
A collection of data objects that are similar to one another within the same group and dissimilar to objects in other groups.
A set of data objects with no similarities or differences.
A group of data objects classified by predefined classes.
A random selection of data objects from a dataset.
Which requirement is NOT typically associated with clustering in data mining?
Scalability
Ability to deal with different types of attributes
Discovery of clusters with arbitrary shape
Predefined class labels for each cluster
A good clustering method should produce clusters with which of the following characteristics?
High intra-class similarity and low inter-class similarity
Low intra-class similarity and high inter-class similarity
Random similarity between clusters
No similarity within or between clusters
Why is the ability to deal with noise and outliers important in clustering methods?
It ensures that clusters are not affected by irrelevant or extreme data points, leading to more accurate results.
It allows clusters to be formed only from outliers.
It makes clustering faster but less reliable.
It is not important for clustering methods.
Which application of cluster analysis would be most relevant for marketers?
Deriving plant and animal taxonomies.
Discovering distinct groups in their customer base and characterizing customer groups based on purchasing patterns.
Classifying documents on the web for information discovery.
Identifying areas of similar land use in an earth observation database.
Suppose you are given a dataset containing interval-based, categorical, and binary data. What requirement must a clustering algorithm meet to effectively analyze this dataset?
The algorithm should only work with interval-based data.
The algorithm should be able to deal with different kinds of attributes.
The algorithm should ignore categorical data.
The algorithm should convert all data to binary format.
A city planner wants to identify groups of houses in a city according to house type, value, and geographic location. Which data mining function would be most useful for this task?
Classification
Regression
Cluster analysis
Association rule mining
How does cluster analysis serve as a tool in data mining?
It helps in predicting future trends.
It provides insight into the distribution of data to observe characteristics of each cluster.
It only detects outliers in data.
It is used exclusively for image processing.
Which of the following is a key requirement for a clustering algorithm to effectively detect clusters of arbitrary shape?
Ability to deal with noisy data
Discovery of clusters with attribute shape
High dimensionality
Interpretability
Why is it important for a clustering algorithm to handle high dimensionality?
To ensure clusters are always spherical
To interpret results more easily
To handle both low-dimensional and high-dimensional data spaces
To avoid noisy data
Which statement best describes a well-separated cluster?
Each object is closer to the prototype than to any other cluster
Each object is closer to every other object in the cluster than to any object not in the cluster
Each object is equally distant from all other objects
Each object is randomly assigned to a cluster
How does a prototype-based cluster differ from a well-separated cluster?
Prototype-based clusters require all objects to be equally distant from the centroid
Prototype-based clusters use a central prototype to define cluster membership, while well-separated clusters rely on pairwise closeness among all members
Well-separated clusters are always spherical, while prototype-based clusters are not
Prototype-based clusters cannot handle categorical data
Why is interpretability an important characteristic of clustering results?
It ensures clusters are always of equal size
It makes the clustering results understandable and usable
It guarantees the absence of noise in data
It restricts clusters to only low-dimensional data
A clustering algorithm is applied to a dataset with missing and erroneous values. Which characteristic is most critical for the algorithm to produce quality clusters?
High dimensionality
Interpretability
Ability to deal with noisy data
Discovery of clusters with attribute shape
How does the shared-property (conceptual clusters) definition of a cluster differ from the graph-based and density-based definitions?
A. It only considers the physical distance between objects.
B. It focuses on objects sharing some property, encompassing previous definitions.
C. It ignores the connections among objects.
D. It requires clusters to be irregular or intertwined.
Suppose you are given a dataset with irregular and intertwined clusters, as well as noise and outliers. Which clustering approach would be most suitable and why?
A. Graph-based, because it connects objects within a specified distance.
B. Density-based, because it can handle irregular clusters and noise.
C. Shared-property, because it groups objects by common properties.
D. None, because clustering is not possible with noise.
In which type of clustering is each point closer to the center of its own cluster than to the center of any other cluster?
Center-based clusters
Contiguity-based clusters
Density-based clusters
Conceptual clusters
A researcher is analyzing a dataset where each point in a cluster is closer to at least one other point in the same cluster than to any point in another cluster. Which clustering method should the researcher use?
Contiguity-based clusters
Center-based clusters
Density-based clusters
Well-separated clusters
Which of the following best describes the main purpose of partitioning methods in clustering?
To classify information into multiple groups based on characteristics and similarity of the data.
To sort data in ascending order.
To encrypt data for security.
To compress data for storage.
According to the partitioning method, what is a requirement for each group formed?
Each group must contain at least one object.
Each group must contain only one object.
Each group must be empty.
Each group must overlap with another group.
Suppose you are given a dataset with 100 objects and you want to use the partitioning method to create 5 clusters. What is the first step according to the K-means algorithm?
Select 5 points as initial centroids.
Sort the data in ascending order.
Remove outliers from the dataset.
Calculate the median of the dataset.
How does the K-means algorithm ensure that the similarity among data objects within a cluster is high?
By assigning objects to the cluster with the nearest mean value.
By randomly assigning objects to clusters.
By grouping objects with different characteristics together.
By minimizing the number of clusters.
Which of the following best describes the stopping criterion for the K-means algorithm?
When the centroids do not change anymore.
When all points are assigned to a single cluster.
When the number of clusters equals the number of points.
When the centroids are at the origin.
Given a dataset with N objects and a specified number K, what does the K-means algorithm output?
A single cluster containing all objects.
A dataset of K clusters.
The sum of all data points.
The minimum and maximum values of the dataset.
Why does the K-means algorithm typically converge to a solution?
Because it always finds the global minimum.
Because points stop shifting from one cluster to another and centroids stop changing.
Because it merges all clusters into one.
Because it removes all outliers.
In the context of K-means clustering, what is the role of the centroid?
It is the point that is farthest from all other points in the cluster.
It is the mean position of all the points assigned to the cluster.
It is the first point assigned to the cluster.
It is the point with the highest value in the dataset.
Which of the following best describes the main purpose of the K-means clustering algorithm as illustrated in the flowchart?
To sort data in ascending order
To group similar data points into clusters based on minimum distance to centroids
To calculate the average of all data points
To remove outliers from a dataset
Suppose you are given a new dataset and asked to perform K-means clustering with K=3. What strategic step should you take first according to the flowchart?
Calculate the distance between all points
Randomly select three initial centroids from the dataset
Assign all points to one cluster
Sort the dataset in ascending order
Which approach is recommended to improve the quality of hierarchical clustering?
Ignore object linkages at each partitioning.
Perform careful analysis of object linkages at each hierarchical partitioning.
Use only non-hierarchical clustering algorithms.
Avoid integrating micro-clusters and macro-clusters.
Why are merging decisions in agglomerative hierarchical clustering considered final?
Because clusters can be split again after merging.
Because the algorithm allows for re-evaluation of merged clusters.
Because once two clusters are merged, the decision cannot be undone.
Because the algorithm optimizes globally after each merge.
Which approach in agglomerative hierarchical clustering gives different weights to points in different clusters based on cluster size?
Unweighted approach
Weighted approach
Random approach
Divisive approach
Suppose you are using agglomerative hierarchical clustering on a dataset with clusters of very different sizes. What strategic consideration should you make when deciding how to merge clusters?
Always merge the largest clusters first.
Treat all clusters as having the same size, regardless of actual size.
Decide whether to use a weighted or unweighted approach for merging.
Only merge clusters with the same number of points.
Given the steps of the basic agglomerative hierarchical clustering algorithm, which step ensures that the algorithm adapts to the new structure after each merge?
Computing the initial proximity matrix
Merging the closest two clusters
Updating the proximity matrix to reflect the new cluster
Stopping when only one cluster remains
Why is the divisive approach considered rigid in clustering?
A. Because clusters can be merged and split multiple times.
B. Because once a merging or splitting is done, it cannot be undone.
C. Because it allows for flexible cluster formation.
D. Because it uses random assignment of clusters.
In divisive hierarchical clustering, what is the main task at each step of the algorithm?
A. Merging the most similar clusters.
B. Splitting a cluster until only singleton clusters remain.
C. Assigning random clusters to objects.
D. Predicting future data trends.
How does divisive clustering differ from agglomerative clustering?
A. Divisive clustering starts with each object in its own cluster.
B. Divisive clustering merges clusters based on similarity.
C. Divisive clustering starts with all objects in one cluster and splits them.
D. Divisive clustering uses prediction models.
What is the primary difference between classification and prediction models in data analysis?
A. Classification models predict continuous values, while prediction models predict categorical labels.
B. Classification models predict categorical labels, while prediction models predict continuous values.
C. Both models predict categorical labels.
D. Both models predict continuous values.
Suppose you are tasked with categorizing bank loan applications as either safe or risky. Which type of model would you use?
A. Prediction model
B. Regression model
C. Classification model
D. Clustering model
Given the evidence that divisive algorithms produce more accurate hierarchies than agglomerative algorithms in some circumstances, what might be a strategic reason to choose divisive clustering for a complex dataset?
A. It is conceptually simpler.
B. It always produces fewer clusters.
C. It can provide more accurate hierarchical structures.
D. It requires less reasoning and planning.
A marketing manager wants to predict how much a customer will spend during a sale. What type of data analysis task is this, and what is the model expected to predict?
Classification; categorical label
Numeric prediction; continuous value
Clustering; group membership
Association; frequent itemsets
During the learning phase of building a classifier, what is the classifier built from?
Randomly generated data points
The training set made up of database tuples and their associated class labels
Only the test data
The results of previous predictions
What is the primary purpose of using a classifier in the classification process?
To generate new data tuples
To estimate the accuracy of classification rules and apply them to new data
To clean the data before analysis
To visualize the data distribution
Which of the following best describes the activity of data cleaning in the context of classification and prediction?
Removing irrelevant features from the dataset
Removing noise and treating missing values using smoothing techniques
Splitting the data into training and test sets
Applying classification rules to the test data
Given the classification rule "IF age=middle-aged AND income=low THEN loan_decision=risky", what would be the loan decision for a person who is middle-aged with low income?
Safe
Risky
Not enough information
High income
Why is preparing the data considered a major issue in classification and prediction tasks?
Because it is the only step that requires human intervention
Because it directly affects the accuracy and reliability of the classification results
Because it is not necessary for most algorithms
Because it is only required for test data
Which of the following best describes the purpose of relevance analysis in a database?
To replace missing values with the most common value
To determine whether any two given attributes are related
To scale all values for a given attribute within a small range
To generalize data to a higher concept
What is the main goal of normalization in data transformation?
To remove irrelevant attributes from the database
To scale all values for a given attribute within a small specified range
To generalize data to a higher concept
To cluster data into groups
Suppose you want to evaluate the effectiveness of a company training program by measuring employee performance before and after the program. Which statistical test would be most appropriate?
Independent sample t-test
Paired sample t-test
ANOVA
Chi-square test
A researcher wants to determine if there is a significant difference in smokers' daily cigarette consumption after using nicotine patches compared to placebo patches. Which statistical method should be used and why?
Independent sample t-test, because the groups are unrelated
Paired sample t-test, because the same subjects are measured twice under different conditions
ANOVA, because there are more than two groups
Regression analysis, because it predicts future values
Why is it important for the independent variable in a paired t-test to consist of two categorical, "related groups" or "matched pairs"?
Because it allows for random assignment of subjects to groups.
Because it ensures that the same subjects are present in both groups, allowing for direct comparison of changes.
Because it increases the number of variables being tested.
Because it eliminates the need for statistical analysis.
A researcher measures participants' typing speed before and after a touch-typing course. What assumption about the independent variable is being met in this scenario?
The independent variable is continuous.
The independent variable consists of two related groups measured at two time points.
The independent variable is random.
The independent variable is binary.
If there are significant outliers in the differences between two related groups in a paired t-test, what is the likely consequence?
The results will be more accurate.
The results may not be valid.
The results will be unaffected.
The results will show a stronger relationship.
Suppose you are planning a study to compare the effectiveness of two treatments for depression using a paired t-test. What reasoning should you use to ensure the validity of your results?
Ensure that the dependent variable is categorical and the groups are unrelated.
Ensure that the dependent variable is continuous, the groups are related, and there are no significant outliers in the differences.
Ensure that the independent variable is continuous and the dependent variable is binary.
Ensure that the groups are randomly assigned and the dependent variable is nominal.
Why can outliers negatively affect the results of a paired t-test?
They distort the differences between the two related groups, reducing accuracy and statistical significance.
They always improve the accuracy of the test.
They make the test results more reliable.
They have no effect on the statistical significance of the test.
Which assumption is required for the distribution of the differences in the dependent variable between two related groups in a paired t-test?
The differences should be approximately normally distributed.
The differences should be exactly equal.
The differences should be completely random.
The differences should be positively skewed.
A researcher finds that the distribution of differences in their paired t-test is not perfectly normal but only slightly violated. What should they conclude about the validity of their results?
The paired t-test is robust to slight violations of normality and can still provide valid results.
The results are completely invalid and must be discarded.
The paired t-test requires perfect normality for valid results.
The results will always be inaccurate if normality is violated.
How can you test for normality in the distribution of differences for a paired t-test using statistical software?
By using the Shapiro-Wilk test of normality.
By calculating the mean and median.
By plotting a bar chart.
By using a paired t-test directly without any checks.
A data scientist is working with a noisy dataset and needs a method that can still make correct predictions. Which criterion should they prioritize when choosing a classification or prediction method?
Speed
Robustness
Interpretability
Scalability
Suppose you are tasked with analyzing a very large dataset and need to choose a classification or prediction method. Which criterion should you focus on to ensure efficient processing?
Accuracy
Interpretability
Scalability
Robustness
