Scribd
Upload
23 views11 pages

Data Mining and Clustering Techniques

Uploaded by

gptmeenakship
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
23 views11 pages

Data Mining and Clustering Techniques

Uploaded by

gptmeenakship
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

DATA MINING

1. What is data mining?


a) Deleting unnecessary data
b) Sorting data alphabetically
c) Storing data securely
d) Extracting useful patterns or information from large datasets
2. Which of the following is not a basic data mining task?
a) Spooling
b) Prediction
c) Classification
d) Clustering
3. Which of the following is not an issue in data mining?
a) High dimensionality
b) Shortage of data
c) Overfitting
d) Outliers
4. Which of the following is not a motivating factor for the development of data
mining tools?
a) Data tombs
b) Data rich but information poor situation
c) Data cleaning
d) Dependency on domain experts in expert systems
5. Which of the following is a subset of data warehouse focused on a specific
functional area?
a) Data mart
b) Association rules
c) Flat files
d) Database
6. Which field of data mining helps in removing uncertainty, noise etc?
a) Data preprocessing
b) Data Mining
c) Outlier detection and removal
d) Uncertainty Reasoning
7. Which among the following are not among Various Operations in Data
Warehousing?
a) Sticking
b) Dice
c) Drill down
d) Roll up
8. Pick the wrong data mining functionality among the given data mining
functionalities.
a) Classification
b) Clustering
c) Class Description
d) Object Description
9. Which of the following is the most effective measure of the center of symmetric
data set?
a) Mode
b) Midrange
c) Mean
d) Median

Data Preprocessing
1. Which of the following is not important in determining data quality?
a) Accuracy
b) Consistency
c) Completeness
d) Database

2. Sometimes the users submit incorrect data in the compulsory fields to avoid divulging
personal information. This is known as _____
a) Disguised missing data
b) Organized missing data
c) Characteristic missing data
d) Coordinated missing data

3. Data cleaning involves filling in missing values.


a) True
b) False

4. Data integration is not a step in data preprocessing.


a) True
b) False

5. Which of the following is not true about data reduction?


a) Reduced data strives to gives same analytical results as the original data
b) Reduced data gives strives to give less accurate analytical results the original
data
c) It involves dimensionality reduction
d) It involves numerosity reduction

6. Which of the following is not a form of data transformation?


a) Normalization
b) Discretization
c) Concept hierarchy
d) Compression

Data Cleaning and Data Integration

1. Which of the following is method is can be used when the class label is missing for a
tuple in the dataset?
a) Ignoring the tuple
b) Generating a copy of the tuple
c) Deleting the label attribute
d) Reversing the tuple

2. Which of the following is a drawback in manual filling of missing values?


a) Incorrect data
b) Redundancy in data
c) Complex data
d) Too many missing values in the data

3. Which of the following is a drawback of filling in a global constant for the missing
value?
a) It increases the data size
b) It decreases the number of missing values
c) It may project wrong trend in data
d) It is difficult to update the data

4. Which of the following is an appropriate measure of central tendency to use fill in


missing values of an attribute of a skewed distribution?
a) Mean
b) Weighted mean
c) Median
d) Geometric mean

5. Which of the following techniques is often used to generate the most probable value
to fill in the missing value?
a) Spooling
b) Decision tree
c) Data cleaning
d) Numerosity reduction
6. Which of the following values are considered in binning for noise smoothing for a data
value?
a) Neighborhood
b) Average
c) Farthest
d) Compressed
7. In smoothing by bin means, each value in the bin is replaced by _____
a) Mean of the values in the bin
b) Median of the values in the bin
c) Mode of the values in the bin
d) Deviation of the values in the bin
8. Data smoothing can also be performed by regression.
a) True
b) False
[Link] of the following is a cause of discrepancy in data?
a) Too many optional fields in data forms
b) Large size of data
c) Redundancy in data
d) Complex data

[Link] of the following is a factor in data discrepancy?


a) Redundant tuple
b) Reversed data
c) Data decay
d) Copied tuple
11. Which of the following is not a factor in data discrepancy?
a) Inconsistent data representation
b) Too many attributes
c) Human errors
d) Data integration
12. The data which informs about the properties about data is known as _____
a) Meta data
b) Data attributes
c) Data codes
d) Data factors
13. Data scrubbing tools do not _____
a) Use domain knowledge
b) Detect errors
c) Make correction in the data
d) Decrease the size of the metadata
14. Data auditing tools do not _____
a) Discover rules and relationships in the data
b) Detect data that violate certain rules
c) Use statistical analysis to find rules in the data
d) Use parsing to find rules in the data
15. The null rule for an attribute specifies _____
a) The data values should not be null
b) All the data values should be null
c) The conditions and handling of null values
d) At most one null value is allowed
16. ETL (Extraction, Transformation, Loading) tools _____
a) Allow the use of graphical user interface
b) Do not allow the use of graphical user interface
c) Do not allow data transformations
d) Are not used for data transformations
17. Which of the following is the chi square value for the given data?

Likes
Likes Cricket Total
Football
Male 30 20 50

Female 40 10 50

Total 70 30 100
a) 2.2
b) 3.7
c) 5.3
d) 4.8

CLUSTERING TECHNIQUES

Got it 👍
Here’s a set of Multiple-Choice Questions (MCQs) on Clustering in Machine Learning
and its Applications.

MCQs on Clustering in Machine Learning


Basic Concepts

1. Clustering in machine learning is mainly a type of:


a) Supervised learning
b) Unsupervised learning
c) Reinforcement learning
d) Semi-supervised learning
Answer: b) Unsupervised learning
2. Which of the following is the main goal of clustering?
a) Predict future values
b) Find decision boundaries
c) Group similar data points together
d) Classify data into known categories
Answer: c) Group similar data points together
3. In k-means clustering, "k" represents:
a) Number of features
b) Number of data points
c) Number of clusters
d) Number of iterations
Answer: c) Number of clusters
4. Which of the following algorithms does not belong to clustering?
a) K-means
b) DBSCAN
c) Agglomerative Hierarchical clustering
d) Naive Bayes
Answer: d) Naive Bayes
Applications

5. Clustering can be used in which of the following real-world applications?


a) Market segmentation
b) Document or news article grouping
c) Image compression
d) All of the above
Answer: d) All of the above
6. In customer segmentation, clustering helps to:
a) Predict exact customer purchases
b) Identify groups of similar customers
c) Rank customers in order
d) Eliminate outliers only
Answer: b) Identify groups of similar customers
7. In biology, clustering can be applied for:
a) Grouping genes with similar expression patterns
b) Predicting protein structures
c) DNA sequence alignment
d) Detecting mutations directly
Answer: a) Grouping genes with similar expression patterns

Algorithm Details

8. The K-means algorithm uses which distance measure most commonly?


a) Manhattan distance
b) Euclidean distance
c) Cosine similarity
d) Hamming distance
Answer: b) Euclidean distance
9. Which clustering algorithm is best suited for arbitrarily shaped clusters and noise
handling?
a) K-means
b) Hierarchical clustering
c) DBSCAN
d) Gaussian Mixture Model
Answer: c) DBSCAN
10. The hierarchical clustering dendrogram is used to:
a) Visualize classification accuracy
b) Show how clusters are merged step by step
c) Represent the centroids in k-means
d) Display probability distributions
Answer: b) Show how clusters are merged step by step

11. Which metric is often used to evaluate clustering quality?


a) Confusion matrix
b) Silhouette score
c) Mean squared error
d) ROC curve
Answer: b) Silhouette score
12. A major drawback of k-means clustering is:
a) It can only handle categorical data
b) It requires the number of clusters (k) to be specified in advance
c) It cannot be applied to large datasets
d) It does not use distance measures
Answer: b) It requires the number of clusters (k) to be specified in advance

MCQs on K-Means Clustering

1. K-means clustering is an example of:


a) Supervised learning
b) Unsupervised learning
c) Reinforcement learning
d) Semi-supervised learning
Answer: b) Unsupervised learning
2. In K-means, “K” represents:
a) Number of features
b) Number of clusters
c) Number of iterations
d) Number of centroids per cluster
Answer: b) Number of clusters
3. Which distance measure is most commonly used in K-means?
a) Manhattan distance
b) Euclidean distance
c) Cosine similarity
d) Hamming distance
Answer: b) Euclidean distance

4. What does the K-means algorithm try to minimize?


a) Sum of squared errors within clusters
b) Maximum distance between clusters
c) Number of iterations
d) Classification error rate
Answer: a) Sum of squared errors within clusters
5. The first step in the K-means algorithm is:
a) Calculate cluster centroids
b) Randomly initialize cluster centroids
c) Assign points to nearest clusters
d) Minimize distance between clusters
Answer: b) Randomly initialize cluster centroids
6. After assigning points to clusters, what happens next in K-means?
a) Centroids are fixed
b) Centroids are recalculated as the mean of assigned points
c) Centroids are deleted
d) Clusters are merged
Answer: b) Centroids are recalculated as the mean of assigned points
7. When does the K-means algorithm stop?
a) When centroids no longer change significantly
b) After exactly 10 iterations
c) When all data points are used once
d) When clusters overlapA major drawback of K-means is:
a) It cannot be applied to numerical data
b) It requires the number of clusters (K) to be chosen in advance
c) It always gives the same result
d) It cannot handle large datasets
Answer: b) It requires the number of clusters (K) to be chosen in advance

8. K-means clustering works best when clusters are:


a) Arbitrary in shape
b) Spherical and equally sized
c) Overlapping and imbalanced
d) Hierarchical in structure
Answer: b) Spherical and equally sized
9. Which method is commonly used to determine the optimal value of K in K-means?
a) Silhouette method
b) Elbow method
c) Cross-validation
d) ROC curve
Answer: b) Elbow method

11. K-means can be applied in:


a) Market segmentation
b) Image compression
c) Document clustering
d) All of the above
Answer: d) All of the above
12. In image compression using K-means, cluster centroids represent:
a) Pixels
b) Color values
c) File size
d) Noise
Answer: b) Color values

MCQs on Hierarchical Clustering


1. Hierarchical clustering is an example of:
a) Supervised learning
b) Unsupervised learning
c) Reinforcement learning
d) Semi-supervised learning
Answer: b) Unsupervised learning
2. Which of the following is not a type of hierarchical clustering?
a) Agglomerative
b) Divisive
c) K-means
d) Both (a) and (b)
Answer: c) K-means
3. In hierarchical clustering, the result is often visualized using:
a) ROC curve
b) Confusion matrix
c) Dendrogram
d) Silhouette plot
Answer: c) Dendrogram

4. Agglomerative hierarchical clustering is also called:


a) Bottom-up approach
b) Top-down approach
c) Random partitioning
d) Centroid initialization
Answer: a) Bottom-up approach
5. Divisive hierarchical clustering is also called:
a) Bottom-up approach
b) Top-down approach
c) K-means extension
d) Distance minimization
Answer: b) Top-down approach

6. A key advantage of hierarchical clustering over K-means is:


a) It works only for large datasets
b) The number of clusters need not be specified in advance
c) It does not require distance measures
d) It always produces spherical clusters
Answer: b) The number of clusters need not be specified in advance
7. A major drawback of hierarchical clustering is:
a) It cannot be used with numerical data
b) It is computationally expensive for large datasets
c) It requires pre-defined cluster centers
d) It does not produce interpretable results
Answer: b) It is computationally expensive for large datasets
8. Hierarchical clustering is commonly used in:
a) Gene expression analysis
b) Social network analysis
c) Document clustering
d) All of the above
Answer: d) All of the above

MCQs on Partitioning vs. Hierarchical


Clustering
Basic Concepts

1. Partitioning clustering methods include:


a) K-means, K-medoids
b) Agglomerative, Divisive
c) DBSCAN, OPTICS
d) PCA, LDA
Answer: a) K-means, K-medoids
2. Hierarchical clustering methods include:
a) K-means and K-medoids
b) Agglomerative and Divisive
c) Naïve Bayes and Decision Trees
d) Logistic Regression and SVM
Answer: b) Agglomerative and Divisive
3. Which of the following is not a partitioning method?
a) K-means
b) K-medoids
c) Agglomerative clustering
d) CLARANS
Answer: c) Agglomerative clustering

Approach & Working

4. Partitioning methods are most suitable for:


a) Small datasets
b) Very large datasets
c) Only categorical data
d) Streaming data only
Answer: b) Very large datasets
5. Hierarchical clustering is most suitable for:
a) Small to medium datasets
b) Extremely large datasets
c) Only time-series data
d) Data with no similarity measure
Answer: a) Small to medium datasets
6. In partitioning methods, the number of clusters (K) must be:
a) Automatically chosen
b) Specified in advance
c) Determined by dendrogram
d) Equal to number of features
Answer: b) Specified in advance
7. In hierarchical clustering, the number of clusters can be:
a) Fixed in advance only
b) Decided by cutting the dendrogram at a certain level
c) Always equal to 2
d) Automatically optimized
Answer: b) Decided by cutting the dendrogram at a certain level
Evaluation & Complexity

8. Partitioning methods are generally:


a) More computationally efficient than hierarchical methods
b) Less efficient than hierarchical methods
c) Equally efficient as hierarchical methods
d) Only applicable to categorical data
Answer: a) More computationally efficient than hierarchical methods
9. A drawback of partitioning methods like K-means is:
a) They cannot handle large datasets
b) They require the number of clusters in advance
c) They cannot use distance measures
d) They produce dendrograms
Answer: b) They require the number of clusters in advance
10. A drawback of hierarchical clustering is:
a) It cannot work without cluster centers
b) It is computationally expensive for large datasets
c) It requires a centroid update step
d) It cannot be visualized
Answer: b) It is computationally expensive for large datasets

11. Partitioning methods are widely used for:


a) Market segmentation in large customer datasets
b) Gene expression analysis (small data)
c) Decision boundary classification
d) Regression tasks
Answer: a) Market segmentation in large customer datasets
12. Hierarchical clustering is widely used for:
a) Image compression
b) Gene expression and biological taxonomy
c) Predicting numerical outputs
d) Time-series forecasting
Answer: b) Gene expression and biological taxonomy

Common questions

Powered by AI

Data mining focuses on extracting useful patterns or information from large datasets. Data preprocessing, a step in data mining, involves preparing data for analysis by cleaning and integrating data, such as filling in missing values and reducing data dimensionality . Data cleaning involves ensuring data quality through consistency and completeness checks. Data warehousing is primarily about storing data in a standardized form for easy access and analysis, while its subset, like a data mart, focuses on specific areas .

Hierarchical clustering is preferred in scenarios requiring visualization of data relationships, such as gene expression analysis, due to its dendrogram output, which illustrates cluster formation steps. It's also beneficial when the number of clusters is unknown, as it does not require a priori cluster number specification . However, it is computationally intensive, limiting its use with very large datasets .

Selecting the appropriate measure of central tendency is crucial in preserving data integrity and avoiding distortion. For skewed distributions, the median is preferred over the mean because it is less influenced by outliers and skewness, providing a more accurate central value for imputation . This choice helps maintain the original data's relative accuracy, which is particularly important in predictive modeling and statistical analyses .

In data warehousing architecture, data marts serve as smaller subsets focusing on specific functional areas or departments, facilitating targeted data analysis. They streamline data mining processes by providing highly relevant and specialized data, enhancing efficiency and reducing processing overload on the larger data warehouse . This focused approach allows for quicker retrieval and analysis, leading to more effective and timely insights .

Clustering is an unsupervised learning method aimed at grouping similar data points together, as seen in k-means or hierarchical clustering techniques . Classification, on the other hand, is a supervised learning task that assigns data points to predefined categories or labels based on training data. The differences imply that clustering is used when the category labels are unknown (e.g., customer segmentation or gene grouping), whereas classification requires predefined labels for predictive tasks (e.g., classifying emails as spam or not).

K-means clustering assumes clusters are spherical and equally sized, which limits its effectiveness on arbitrarily shaped clusters and datasets with noise. K-means also requires specifying the number of clusters (k) in advance . DBSCAN overcomes these limitations by identifying clusters based on density, allowing for non-spherical shapes and better handling of noise. It dynamically detects the number of clusters, making it more adaptable to complex data structures .

High dimensionality in data mining can lead to the 'curse of dimensionality,' where the data becomes sparse, and computational efficiency decreases due to the large number of features. It complicates the pattern recognition and increases risks like overfitting. Mitigation strategies include dimensionality reduction techniques such as principal component analysis (PCA), feature selection methods, and data preprocessing techniques like data reduction .

The elbow method determines the optimal number of clusters by plotting the variance explained as a function of the number of clusters, identifying a point where adding more clusters yields diminishing improvements in variance explained. Despite its utility, the method can be subjective as the elbow point is not always distinct, especially in data without a clear cluster structure .

Metadata, which provides information about data properties, plays a critical role in ensuring data quality during preprocessing and integration by detailing the data's origins, context, structure, and relationships. This information facilitates accurate data alignment, consistency checks, and integration, enhancing the quality of data cleaning and harmonizing datasets from multiple sources .

Data cleaning ensures accuracy, consistency, and completeness of data, which is crucial for reliable data mining outcomes. Common techniques include filling in missing values, detecting and removing outliers, handling inconsistencies, and ensuring data integration. By addressing data quality issues, data cleaning helps in reducing errors and improving the model's predictive performance .

You might also like