NEW
Font size
S
M
L
XL
WorksheetsWeb Mining 3
Total questions: 76
Worksheet time: 38mins
Name
Class
Date
1.
According to the slides, data objects represent what in the data?
a)
Entities such as customers, products, or transactions
b)
Only numeric attributes
c)
Only database tables
d)
Only images and videos
e)
Only web pages
2.
What is the set of attributes that describe a given object called?
a)
Attribute vector (feature vector)
b)
Histogram bucket
c)
Quantile chart
d)
Projection plane
e)
Decision boundary
3.
Nominal attribute values are valuable primarily as what?
a)
Symbols or names
b)
Precise distances on a scale
c)
Ratios with a true zero
d)
Ordered ranks
e)
Continuous measurements
4.
A binary attribute has how many categories or states?
a)
Two
b)
Three
c)
Four
d)
Five
e)
An unlimited number
5.
Ordinal attribute values follow what property?
a)
A certain order
b)
A true zero that allows multiplication
c)
Equal spacing between values
d)
No meaningful ordering
e)
Only two possible states
6.
Which statement best matches an interval attribute?
a)
Values are measured along an equally spaced scale
b)
Values are only category names
c)
Values have only two states
d)
Values must be unique identifiers
e)
Values cannot be compared
7.
Which statement best matches a ratio attribute in the slides?
a)
A numerical attribute with a value 0 that supports multiplication
b)
A categorical attribute with no ordering
c)
An attribute defined only by symbols
d)
An attribute measured only in ranks
e)
An attribute that cannot take the value 0
8.
An attribute is discrete if it has what kind of set of values?
a)
A finite or countably infinite set
b)
An uncountably infinite set only
c)
Only negative values
d)
Only positive values
e)
Exactly two values
9.
Which item is explicitly listed under basic data statistics in the slides?
a)
Identify outliers
b)
Train a deep neural network
c)
Encrypt the dataset
d)
Build a web crawler
e)
Generate random data
10.
The slides note that the mean is sensitive to what issue?
a)
Outliers
b)
Missing labels
c)
Tree depth
d)
Pixel resolution
e)
Number of neurons
11.
What does the median do according to the slides?
a)
Divides the data into larger and smaller parts with the same number of elements
b)
Finds the most frequent value
c)
Computes max minus min
d)
Averages the maximum and minimum values
e)
Counts how many values are below each point
12.
In the grouped-data median approximation, what does L1 represent?
a)
The lower boundary of the median interval
b)
The upper boundary of the median interval
c)
The maximum value in the dataset
d)
The mean of all values
e)
The mode value
13.
What is the mode of a dataset?
a)
The most frequent value in the dataset
b)
The average of all values
c)
The point that divides the data into two equal parts
d)
The difference between the largest and smallest values
e)
The average of the maximum and minimum values
14.
What is the midrange as defined in the slides?
a)
The average of the maximum and minimum values
b)
The most frequent value
c)
The value that splits data into equal parts
d)
The sum of all values divided by the number of values
e)
The cumulative frequency below each value
15.
What is the range as defined in the slides?
a)
The largest value minus the smallest value
b)
The average of the largest and smallest values
c)
The most frequent value
d)
The median interval width
e)
The standard deviation squared
16.
According to the slides, the interquartile range (IQR) equals what?
a)
Q3 minus Q1
b)
Q1 minus Q3
c)
Median minus mean
d)
Max minus min
e)
Mean minus mode
17.
In a boxplot, Q1 and Q3 correspond to what parts of the box?
a)
The beginning and the end of the box
b)
The minimum and maximum of the dataset
c)
Two different means
d)
Two different medians
e)
Two different modes
18.
Standard deviation is described in the slides as representing what?
a)
Dispersion of the data relative to the mean
b)
The most frequent value
c)
The lower boundary of the median interval
d)
The number of histogram buckets
e)
The number of data dimensions
19.
For a quantile chart, what is the first step described in the slides?
a)
Sort values in ascending order
b)
Sort values in descending order
c)
Convert values to categories
d)
Remove all outliers
e)
Randomly shuffle the values
20.
A quantile-quantile chart is used to show the relationship between what?
a)
Quantile values of two univariate distributions
b)
Two different histogram bucket widths
c)
Pixel colors across windows
d)
Tree nodes in a hierarchy
e)
Neurons and their reference vectors
21.
In a histogram, values are grouped into equal intervals called what?
a)
Buckets
b)
Reference vectors
c)
Quantiles
d)
Dimensions
e)
Modes
22.
A scatter chart is used to determine what between two numeric attributes?
a)
Reciprocity (relationship)
b)
The number of bins needed
c)
The attribute type
d)
The median interval
e)
The tree root
23.
Which term is used in the slides for no reciprocity between two numeric attributes?
a)
Null reciprocity
b)
Positive reciprocity
c)
Negative reciprocity
d)
Weighted reciprocity
e)
Ordinal reciprocity
24.
In pixel-based visualization, a data dimension value is represented by what?
a)
A pixel whose color corresponds to the value
b)
A histogram bucket whose height is the value
c)
A tree node whose depth is the value
d)
A reference vector with a single element
e)
A bar whose width is the value
25.
In pixel-based visualization, records are usually sorted by what?
a)
A dimension of interest
b)
Random order to avoid bias
c)
Only by the median value
d)
Only by the mode value
e)
Only by the maximum value
26.
In pixel-based visualization, how is a larger value typically encoded compared to a smaller value?
a)
A darker color
b)
A lighter color
c)
A larger text label
d)
A longer axis line
e)
A bigger window count
27.
Visualization in vector space is introduced because pixel-based visualization does not represent what?
a)
The density of data points
b)
The number of attributes
c)
The names of attributes
d)
The median of the data
e)
The histogram bucket width
28.
In scatter charts, what can be used to represent a third dimension?
a)
Different shapes
b)
Only the X axis
c)
Only the Y axis
d)
A second histogram
e)
A super sphere tree
29.
The slide labeled 'Scatter Chart (cont)' continues discussion of which chart type?
a)
Scatter chart
b)
Histogram
c)
Boxplot
d)
Quantile chart
e)
Pixel-based visualization
30.
According to the slides, a basic scatter chart can represent up to how many dimensions?
a)
Four
b)
Two
c)
Three
d)
Five
e)
Six
31.
Which dataset is used as an example in the lecture?
a)
Iris dataset
b)
MNIST dataset
c)
CIFAR-10 dataset
d)
IMDB reviews dataset
e)
ImageNet dataset
32.
A super sphere tree is described as being suitable for data that has what structure?
a)
A tree structure
b)
A linear sequence only
c)
A single table with no relationships
d)
A purely random graph only
e)
A fixed 2D grid only
33.
A super sphere tree supports focusing on part of the data while still representing what?
a)
The general context of the data
b)
Only the selected node
c)
Only the median value
d)
Only the histogram buckets
e)
Only the color legend
34.
According to the slides, SOM (Self Organizing Map) is what type of neural network?
a)
A feed-forward neural network with two layers
b)
A recurrent neural network with memory cells
c)
A convolutional neural network with pooling
d)
A transformer with attention heads
e)
A decision tree ensemble
35.
In SOM, the association weights from the input layer to a neuron are called what?
a)
A reference vector
b)
A histogram bucket
c)
A quantile
d)
A feature count
e)
A tree edge
36.
During SOM competitive learning, which neurons have their weights updated?
a)
The winning neuron and its neighboring neurons
b)
All neurons in the competition layer
c)
Only the input layer neurons
d)
Only the farthest neuron from the input
e)
No neurons are updated
37.
In SOM competitive learning, how is the winning neuron selected for an input vector?
a)
It has the smallest distance to the input
b)
It has the largest distance to the input
c)
It has the highest index number
d)
It is chosen randomly
e)
It is always the center neuron
38.
In WEBSOM, texts are represented as what?
a)
Bags of words
b)
Decision trees
c)
Image pixels
d)
Time series signals
e)
Relational database schemas
39.
After learning in WEBSOM, each text group can be represented by what?
a)
Specific keywords
b)
Only numeric IDs
c)
Only colors
d)
Only node sizes
e)
Only page numbers
40.
Hair color is used in the slides as an example of which attribute type?
a)
Nominal attribute
b)
Ordinal attribute
c)
Interval attribute
d)
Ratio attribute
e)
Continuous attribute
41.
Which example from the slides best fits an interval attribute?
a)
Temperature on the Celsius scale
b)
Customer ID
c)
Hair color
d)
Gender
e)
Quantity of items
42.
Which example from the slides best fits a ratio attribute?
a)
Money
b)
Temperature on the Celsius scale
c)
Hair color
d)
Size categories
e)
Customer ID
43.
Size values like small, normal, large, oversized are best classified as what type of attribute?
a)
Ordinal attribute
b)
Nominal attribute
c)
Interval attribute
d)
Ratio attribute
e)
Binary attribute
44.
A dataset where every value is unique is described in the slides as having what property regarding the mode?
a)
It has no mode
b)
It is always multimodal
c)
Its mean equals its mode
d)
Its median cannot be computed
e)
Its range must be zero
45.
If you want a measure of central value that the slides explicitly warn is sensitive to outliers, which measure is that?
a)
Mean
b)
Mode
c)
Median
d)
Midrange
e)
Interquartile range
46.
If some observations should contribute more than others, which mean described in the slides should be used?
a)
Weighted mean
b)
Median
c)
Mode
d)
Midrange
e)
Range
47.
In the grouped-data approach to estimating the median, what step comes before locating the median interval?
a)
Calculate the frequency of values in each interval
b)
Compute the range as max minus min
c)
Compute the interquartile range
d)
Convert the values into categories
e)
Select the most frequent value
48.
Which pair of values defines the endpoints of the box in a boxplot, according to the slides?
a)
Q1 and Q3
b)
Min and max
c)
Mean and median
d)
Mode and median
e)
Standard deviation and variance
49.
Which statement is consistent with the slides' definition of quantiles?
a)
Quantiles divide data into nearly equal parts with equal numbers of elements
b)
Quantiles divide data into equal-width numeric intervals
c)
Quantiles are the most frequent values in the dataset
d)
Quantiles measure dispersion relative to the mean
e)
Quantiles are only defined for binary attributes
50.
If you want to split a dataset into 100 nearly equal parts as described in the slides, which concept should you use?
a)
Percentiles (100-quantile)
b)
Quartiles (4-quantile)
c)
Median only
d)
Range
e)
Midrange
51.
In a quantile chart, the quantity f_i is described as what?
a)
The percentage of data with values below a given value
b)
The width of a histogram bucket
c)
The maximum value of the dataset
d)
The standard deviation of the dataset
e)
The number of data dimensions
52.
A scatter chart is intended for which type of pair of attributes?
a)
Two numeric attributes
b)
Two nominal attributes
c)
One nominal and one ordinal attribute only
d)
Two binary attributes only
e)
One discrete ID and one nominal attribute only
53.
According to the slides, how can a scatter chart represent four dimensions?
a)
Use X and Y axes, plus shape for the third dimension and color for the fourth
b)
Use four different X axes
c)
Use four different Y axes
d)
Use only color for all dimensions
e)
Use histogram buckets instead of points
54.
If your data has more than four dimensions, which approach do the slides recommend instead of a single scatter chart?
a)
Scatter chart matrix
b)
Boxplot
c)
Histogram
d)
Quantile-quantile chart
e)
Midrange chart
55.
In pixel-based visualization, a data point with m dimensions is represented in how many windows?
a)
m windows
b)
One window
c)
Two windows
d)
Four windows
e)
A variable number independent of m
56.
In pixel-based visualization, why are records often sorted in a dimension of interest?
a)
To make color patterns reveal correlations between dimensions
b)
To reduce the number of dimensions to two
c)
To remove outliers from the dataset
d)
To convert continuous attributes into nominal ones
e)
To ensure the mode equals the median
57.
The slides say vector-space visualization uses projection techniques primarily to do what?
a)
Represent multidimensional data in a 2D space
b)
Compute the standard deviation more accurately
c)
Convert nominal attributes into ratio attributes
d)
Guarantee the data has no outliers
e)
Create a tree structure from the data
58.
A super sphere tree is described as useful when you need to do what at the same time?
a)
Focus on part of the data while still showing overall context
b)
Compute the exact median without sorting
c)
Convert a ratio attribute to an interval attribute
d)
Represent density using pixels only
e)
Compare two distributions using quantiles only
59.
In SOM, the input layer dimension is described as being equal to what?
a)
The dimension of the input data
b)
The number of neurons in the competition layer
c)
The number of histogram buckets
d)
The number of quartiles
e)
The number of tree levels
60.
In SOM competitive learning, what distance is used between an input vector and a neuron?
a)
Euclidean distance between the input and the neuron's reference vector
b)
Hamming distance between two binary codes
c)
Edit distance between two strings
d)
Cosine similarity between two documents
e)
Manhattan distance between two histogram buckets
61.
In WEBSOM, what does a high-density area on the 2D map indicate according to the slides?
a)
Many texts are concentrated in that area
b)
The dataset has no outliers
c)
The mean equals the median
d)
The attribute is continuous rather than discrete
e)
The data has exactly four dimensions
62.
You suspect your numeric dataset contains extreme outliers and you need a central value. Based on the slides, which choice best avoids the outlier sensitivity explicitly attributed to the mean?
a)
Median
b)
Mean
c)
Midrange
d)
Range
e)
Histogram bucket width
63.
You want to estimate the median from grouped data using the method described in the slides. Which piece of information is NOT part of the median-approximation components listed?
a)
Maximum value of the dataset
b)
Lower boundary of the median interval
c)
Number of values in the dataset
d)
Sum of frequencies below the median interval
e)
Width of the median interval
64.
You need a single statistic that summarizes the spread of the middle portion of the data rather than the full min-to-max span. Which quantity defined in the slides best matches that goal?
a)
Interquartile range (IQR)
b)
Range
c)
Midrange
d)
Mode
e)
Mean
65.
You want to compare two univariate distributions using their quantile values rather than raw values. Which visualization described in the slides is designed for that purpose?
a)
Quantile-quantile chart
b)
Quantile chart
c)
Histogram
d)
Boxplot
e)
Scatter chart
66.
A basic scatter chart can represent up to four dimensions, but your dataset has many more. You still want to visualize relationships among all pairs of dimensions. Which approach in the slides best fits?
a)
Scatter chart matrix
b)
Single scatter chart with more colors
c)
Histogram with many buckets
d)
Boxplot for each data point
e)
Midrange plot for each attribute
67.
You need to visualize multidimensional data in 2D and also care about how densely points occupy regions. Which combination of slide claims most directly motivates moving from pixel-based visualization to vector-space visualization?
a)
Pixel-based visualization does not represent point density, so projection to 2D vector space is used
b)
Pixel-based visualization requires Euclidean distance, so projection is avoided
c)
Vector-space visualization is only for binary attributes, so pixels are replaced
d)
Scatter charts cannot show reciprocity, so pixels are used instead
e)
Histograms require tree structure, so vector space is required
68.
In pixel-based visualization, you sort records by one dimension of interest and then inspect colors across multiple windows. What is the intended outcome of this procedure according to the slides?
a)
Correlations between dimensions become visible through color distributions across windows
b)
The data becomes continuous instead of discrete
c)
The mode of each dimension becomes the same
d)
The number of dimensions is reduced automatically
e)
Outliers are removed from the dataset
69.
A visualization method in the slides is intended for large tree-structured data and uses fish-eye behavior where focused items grow while unfocused items shrink. Which method is this?
a)
Super sphere tree
b)
Scatter chart matrix
c)
Quantile chart
d)
Pixel-based visualization
e)
Histogram
70.
Consider SOM training as described: pick the neuron with minimum distance to the input, then update weights for the winner and its neighbors. Which statement best explains why the competition layer has an explicit spatial organization (rectangle, hexagon, etc.)?
a)
Because neighborhood relations determine which additional neurons get updated
b)
Because the input layer must be two-dimensional
c)
Because the mean is computed on a grid
d)
Because quantiles require a geometric layout
e)
Because scatter charts require shapes for the third dimension
71.
A student claims that in SOM competitive learning, updating all neurons would be equivalent to updating only the winner and neighbors. Based on the slide description, which response best matches the algorithm as stated?
a)
No, the slides specify updating only the winning neuron and its neighbors
b)
Yes, the slides specify updating all neurons in every step
c)
Yes, but only when the input is binary
d)
No, because the slides forbid using Euclidean distance
e)
Yes, because the slides use a histogram objective
72.
You are given a dataset with m dimensions and you use pixel-based visualization exactly as described. Which statement must be true about how one data point is displayed?
a)
It appears as m pixels, one in each of m windows
b)
It appears as a single pixel whose color is the average of all dimensions
c)
It appears as one point in a 2D projection without color
d)
It appears only if the data is continuous
e)
It appears only if m is less than or equal to four
73.
Attribute type in the slides is said to be determined by the set of attribute values. Which situation would most directly change an attribute's type under this definition?
a)
Changing the allowed value set from ordered ranks to unordered category names
b)
Reordering the rows of the dataset
c)
Changing the number of objects in the dataset
d)
Changing the color mapping in a pixel-based view
e)
Changing the font used in a chart title
74.
A team wants to use WEBSOM to map documents. They plan to feed full word-order sequences into the model without converting them. Which slide-aligned correction is most appropriate?
a)
Represent texts as bags of words before learning
b)
Convert texts into histogram buckets of equal width
c)
Convert texts into ratio attributes with true zero
d)
Compute the midrange of each document
e)
Sort documents by a dimension of interest to create windows
75.
After WEBSOM training, you see a region with high document density. According to the slides, which additional property is most likely available to characterize groups in that region?
a)
Specific keywords representing the text group
b)
Exact median values for each word
c)
A boxplot showing min and max word lengths
d)
A scatter chart with four numeric dimensions
e)
A tree root representing the hierarchy of documents
76.
You need to choose between range and standard deviation to describe dispersion relative to the mean. Which measure is explicitly described in the slides as relative to the mean?
a)
Standard deviation
b)
Range
c)
Midrange
d)
Mode
e)
Median
Reset
