WorksheetsData Mining
Total questions: 69
Worksheet time: 39mins
The problem of finding hidden structrures in unlabeled data is called:
Supervised learning
Unsupervised learning
Candidate Generation
None of the above
Which of the following is not a process in data mining?
Knowledge extraction
Cluster analysis
Data exploration
Data Transformation
Bayesian classifier is:
A class of learning algorithm that employs posteriori properties in classification of a new record.
Decision Tree Induction
A class of learning algorithm that employs Gain Ratio in classification of a new record.
None of the above
Classification Accuracy is:
A subdivision of a set of examples into a number of classes
Measure of the accuracy of the classification of a concept that is given by a certain theory
The task of assigning classification to a set of examples
None of the above
A cluster is
A group of similar objects that differ significantly from other objects in a different group.
Operation on a database to transform or simplify data in order to prepare it for a machine learning algorithm
Symbolic representation of facts or ideas from which information can potentially be extracted
Assuming log base 2, the entropy of a binary feature with (# of Ys = 5 and # of Ns = 5) in a partition with 10 tuples is:
0
1
0.5
None of the above
What is the equation to calculate Accuracy?
(FP + FN) / (P + N)
(TP) / (P)
(TP + TN) / (P + N)
(TP) / (TP + FP)
Which of the following statements is true assuming the confidence of a 2-itemset frquent pattenr (I1, I2) is 65% and the lift is 1.86%?
I1 is correlated to I2
I2 is correlated to I1
a and b
none of the above
Which of the following is an unsupervised data mining technique?
Cluster analysis
Feature extraction
Frequent pattern mining
All of the above
What is the equation to calculate Precision?
(TN) / (P)
(TP) / (P)
(TP) / (TP + FP)
(TP + TN) / (P + N)
Which of the following methods is the recommended method for estimating accuracy of a classifier?
Holdout
Random
Sampling
Stratified 10-fold cross-validation
5-fold cross validation
Which of the following central tendency measures is used in attribute selection method for defining splitting criterion for an attribute?
Mean of the attribute
Median of the attribute
Mode of the attribute
None of these
What is the equation to calculate the Error rate?
(2 * precision * recall) / (precision + recall)
(TP) / (TP + FP)
(TP + TN) / (P + N)
(FP + FN) / (P + N)
In a decision tree classifier, a pure partition refers to when:
All tuples in the partition belong to different classes
The class label is evenly distributed in the partition tuples
The entropy of an attribute is not known.
None of the above.
In a decision tree classifier, a pure partition refers to when:
All the tuples in the partition belong to the same class
The class label is evenly distributed in the partition tuples
The entropy of the attribute is unknown.
None of these.
What is the equation to calculate the Specificity?
(TN) / (P)
(TP) / (TP + FP)
( (1 + p^2) * precision * recall) / (p^2 * precision + recall) )
(2 * precision * recall) / (precision + recall)
What is the main limitation of the Naive Bayesian classification algorithm?
It assumes that the attributes are independent of each other given the output class.
It requires taking additional measures for outcomes of an attribute with zero probability
It makes it hard to deal with continuous number attribute types.
All of these.
Using the confusion matrix above, what are the accuracy and the sensitivity of the classifier respectively?"
46+412/7000 and 412/10000
(6954 + 46)/10,000 and 6954/7000
(6954 + 2588)/10,000 and 6954/7000
46+412/10000 and 412/10000
Using the table above, calculate the EXPECTED values for buys_computer_yes, and buys_computer_no.
5156, 1843
2209, 790
5156, 2843
2119, 890
6012, 1843
2209, 790
2156, 4843
1119, 590
What is the Hold Out method?
The given data is randomly partitioned into two independently training and test sets
The accuracy is the overall number of correct classification from the k iterations divided by the total number of tuples in the initial data.
The initial data are randomly partitioned into k mutually exclusive subsets or folds of each approximately equal size. Training and testing is performed k times.
The accuracy is the overall number of correct classification from the k iterations divided by the total number of tuples in the initial data.(edited)
What cross-validation technique is recommended for estimating accuracy?
Stratified 10-fold cross-validation
K-fold cross-validation
Leave-one-out
Leaf node validation
Knowledge Extraction is...
Grouping a set of objects in such a way that objects in the same group are more similar than those in the other groups
an approach where the analyst uses visual exploration to understand the characteristics of the data, rather than through traditional methods
the process of converting your data from one format (or structure) into a different type of format or structure.
the process of creating knowledge from a set of data, such as images or a database. The result needs to be machine-readable so we can use it.
The Bayesian Classifier is a probability classifier. True or False?
True
False
Naive Bayesian benefits from having strong feature independence assumptions
True
False
What is Leave-one-out k-fold cross validation?
K separate times, the data is trained on a single point of data.
N separate times, the data is trained on ALL data except for one point and a prediction is made for that point.
N separate times, the data is trained one by one. Each iteration removed a single training point. When completed, a prediction is made.
K separate times, the data is trained on K-1 times. After N iterations, a prediction is made.
Why do we use Gain Ratio? (Select all that apply)
Overcomes the bias of Information gain.
Applies a kind of normalization to information gain using a split information value.
This algorithm is mostly used in text classification and with problems having multiple classes.
To evaluate the accuracy of our classifier.
K-means clustering is
K-means clustering is a type of unsupervised learning that operates on unlabeled data
K-means clustering is a type of supervised learning that operates on unlabeled data
K-means clustering is a type of unsupervised learning that operates on labeled data
None of the above.
The K-medoids algorithm ...
The K-medoids algorithm breaks up our data in groups and attempt to minimize the distance between "points"
The K-medoids algorithm also calculates the lift for each tuple for future predictions.
The K-medoids algorithm runs K-1 times, calculating the information gain for each tuple.
None of the above.
What are the partitioning methods of clustering? (Select all that apply)
Partitioning Method
Standard Method
Density-Based Method
Grid-Based Method
Time-Based Method
What are the 4 metrics for evaluation classifier performance?
Accuracy, Precision, Sensitivity, Confusion Matrix
Information Gain, Precision, Lift, Chi Square
Chi Square, Lift, Precision, Ratio Gain
Accuracy, Sensitivity, Lift, Confusion Matrix
Bayes theorem converts the results from your test into the real probability of the event. True or False?
True
False
What 3 things are used to validate minimum confidence.
Information Gain
Lift
Chi Square
Ratio Gain
Kulcyznski
Which of the following isn't a Supervised Learning process?
Classification
Regression
Clustering*
Anomaly Detection
Feature Extraction is a type of Supervised Learning.
True
False
Which of the following calculates Classfication Accuracy?
Correct + Negative Prediction / Total Prediction
Total Prediction - Negative Prediction / Correct Prediction
Correct Prediction / Total Prediction
Negative Prediction / Correct Prediction
How do you determine the root of a Decision Tree?
SplitInfo(Class) - SplitInfo(Column)
InformationGain(Column)- SplitInfo(Class)
InformationGain(Class) - InformationGain(Column)
SplitInfo(Column) - InformationGain(Class)
When is a partition considered "Pure" in a Decision Tree?
When the values of the class column are identical.
When all row and columns have the same values.
When the values of any non-class column are identical.
When the values of any row has the same values.
When determining Accuracy in a Confusion Matrix Table, we add True Positive, True Negative together and divide by adding False Negative and False Positive.
True
False
What is a false limitation of the Bayesian Classification when comparing it to Decision Tree induction:
Requires each conditional probability to be a non-zero value.
Any dependencies cannot be modeled by it.
A category that wasn't observed assigns a null probability to it.
It uses the Laplacian Correction method to correct non-zero values.
Supervised Learning is done when... (Select all that apply)
have a specific target value we want to predict about our data.
are uncertain of the target value we want to predict about our data
We want to identify new data points that match our target value.
None of the above
Select all the fields that are relevant to supervised learning.
(Select all that apply)
Classification
Clustering
Regression
Anomaly Detection
Association
Select all the fields associated with unsupervised learning.
(Select all that apply)
Clustering
Association
Feature Extraction
Anomaly Detection
Regression
A confusion matrix is a technique for summarizing the performance of a Classification Algorithm.
True
False
Classification accuracy is the ratio of correct prediction to total predictions made multiplied by all incorrect predictions.
True
False
Classification accuracy is the ratio of correct prediction to total predictions made multiplied by 100
True
False
A Clustering Algorithm finds and groups data points with natural similarities.
True
False
The higher the Lift, the lower the correlation it has.
True
False
The higher the Entropy, the better the predictions we can do.
True
False
Which of the following data warehouse models could be in a more normalized state than the others?
Star Schema
Fact Constellation
Snowflake Schema
(a) and (b)
OLTP stands for:
Online Transformation and Planning
Online Technical Presentation
Online Transaction Processing
None of the above
Which of the following is a characteristic of a Data Warehouse system? (circle all that apply)Subject Oriented
History Data
De-normalized table structure (few tables, many columns per table)
Non-volatile
Continuous updates
Subject Oriented
A table that contains data consisting of numeric measurements of a business process is:
Fact table
Dimension table
Data Dictionary
Metadata
Which of the following is a step in the dimensional design process?
Identify the fact
Declare the grain
Identify the dimensions
Select the business process
All of the above
Which of the following attribute types is appropriate to present the value of a temperature measured as 18 degrees Kelvin?
Nominal
Binary
Ratio Scaled
Interval Scaled
Which of the following numbers are considered as outliers if the Five Number summary for an observation are:
Q1 = 10
Q2 = 14
Q3 = 18
Smallest Observation = 2
Largest Observation = 28
4 and 32 are both outliers
Neither 4 nor 32 are outliers
4 is an outlier but 32 is not
4 is not an outlier but 32 is
Which of the following processes are used to deal with missing data in a dataset?
use a global constant to fill in the missing value
Use a measure of central tendency for the attribute
Use the attribute mean or median for all samples belonging to the same class as the giving tuple
Use linear regression to fill in the missing values
All of the above
Which of the following definitions best describes a closed itemset?
A closed itemset is an itemset that supports min-support threshold.
An itemset X is a closed itemset in a dataset D, if there exists no proper super itemset Y such that Y has the same support count as X in D.
Every frequent itemset is a closed itemset.
Closed itemset is the same as maximal closed itemset.
All of the above.
Normalization is the process of ____________________.
Transforming the data to fall within a smaller or common range such as [-1, 1] or [0.0, 1.0].
replacing missing values
removing outliers
none of the above
Which of the following algorithms are types of unsupervised data mining? (Circle all that apply?
Classification
Associations
Clustering
Feature Extraction
What is the probability of drawing two kings from a deck without repeating?
(2/52)
(4/52)
(4/52) * (4/52)
None of the above
In the concept of frequent pattern mining, which frequent patterns are considered interesting?
The association rule supports minimum support threshold.
The association rule supports minimum confidence threshold.
(a) and (b)
None of the above
Evaluate the following frequent itemsets and their support counts in a transaction dataset. If the min-support = 6, which of the following 3-itemset is NOT a frequent itemset?
I1, I2, I3
I2, I3, I4
I1, I2, I4
None of the above
Suppose that a group of 1000 people were surveyed. Evaluate the following observations from the survey:
Participants who read only fiction books = 250
Participants who read only non-fiction books = 350
Participants interested in reading both fiction and non-fiction books = 150
What is the confidence of a participant (X, "fiction books") --> reads (X, "non-fiction books")?
40%
60%
66%
None of the above
Support that the data for analysis includes the attribute age. The age value for the data tuples are (in increasing order) 14, 16, 18, 18, 20, 21, 22, 22, 23, 24, 26, 33, 34, 35, 36, 38, 42.
What is the mean of the data?
21
24
25
26
Support that the data for analysis includes the attribute age. The age value for the data tuples are (in increasing order) 14, 16, 18, 18, 20, 21, 22, 22, 23, 24, 26, 33, 34, 35, 36, 38, 42.
wHAT IS THE MEDIAN?
22.1
22.5
25
22
Support that the data for analysis includes the attribute age. The age value for the data tuples are (in increasing order) 14, 16, 18, 18, 20, 21, 22, 22, 23, 24, 26, 33, 34, 35, 36, 38, 42.
What is the mode?
18
22
26
33
Support that the data for analysis includes the attribute age. The age value for the data tuples are (in increasing order) 14, 16, 18, 18, 20, 21, 22, 22, 23, 24, 26, 33, 34, 35, 36, 38, 42.
What is the Q1 value?
16
17
16.5
17.5
Task of deriving a model from labeled training data is called:
Supervised learning
Unsupervised learning
Candidate generation
None of the above
The director of customer relationship at All Electronics wants to organize all their customers into 5 groups based on the customers spending habits, so that every group can be assigned to a different manager in the company. This task is called:
Supervised Learning
Data Extraction
Unsupervised Learning
Data Transformation
