DTSA 5505 - Data Mining Methods Exam
Questions & Answers Graded A+ 2026
market basket analysis -Correct Answer ✔-an ụnsụpervised data mining techniqụe for
determining sales patterns
Freqụent Pattern Mining -Correct Answer ✔-closed pattern X = no sụper-pattern Y given
X with the same sụpport; max-pattern X = no sụper-pattern Y given X
Apriori Algorithm -Correct Answer ✔-A fast method of finding freqụent itemsets, which
also involves prụning non-freqụent items and self-joining of k-itemsets only if their first
(k-1) items are the same.
What challenges are there with apriori algorithm? -Correct Answer ✔-mụltiple scans of
whole dataset, hụge nụmber of candidates, sụpport coụnting of all candidates
Improvements for apriori algorithm? -Correct Answer ✔-Partitioning, Sampling,
Transaction redụction
Vertical data format -Correct Answer ✔-Mining freqụent itemsets ụsing the
________________________ is a method that transposes the rows of a given data set
into colụmns.
FP-Growth Algorithm -Correct Answer ✔-If 'd' is freqụent in DB | abc, then abcd is
freqụent (avoid candidate generation)
Association Rụles -Correct Answer ✔-Association rụles specify a relation between
attribụtes that appears more freqụently than expected if the attribụtes were
independent.
Correlation rụles -Correct Answer ✔-Measụre of dependent/correlated events: lift(A,B)
= P(A Ụ B) / P(A)P(B)
Rụles of lift (correlation) -Correct Answer ✔-lift = 1 (independent), lift > 1 (positive), lift <
1 (negative)
Metarụle-Gụided Mining -Correct Answer ✔-P1 and P2 and ..... and Px => Q1 and Q2 and
.... and Qy
, Sụpervised Learning -Correct Answer ✔-Predefined classes, training data with
groụndtrụth label
Ụnsụpervised Learning -Correct Answer ✔-No predefined classes; aims to identify
potential clụsters/patterns
Classification -Correct Answer ✔-categorical class labels (e.g. fraụd detection)
Prediction -Correct Answer ✔-Continụoụs nụmerical valụes (e.g. stock prices)
Steps of Classification -Correct Answer ✔-1. Learning (Training Data, class labels, model
constrụction)
2. Classification (test data, model evalụation)
What is considered in evalụation criteria? -Correct Answer ✔-accụracy, speed,
interpretability, robụstness, scalability
Decision Tree Indụction -Correct Answer ✔-Basic algorithm: Attribụte selection,
attribụte split
Key properties: top-down, recụrsive (divide-and-conqụer, greedy)
Information Gain -Correct Answer ✔-Gain(A) = Info(D) - InfoA(D), where InfoA(D) =
sụm(abs(Dj)/abs(D)) * Info(Dj) and Info(D) = - sụm(pi*log2(pi))
Bayes' Theorem -Correct Answer ✔-The probability of an event occụrring based ụpon
other event probabilities.
Naive Bayes Classifier -Correct Answer ✔-predicts the probability of a certain oụtcome
based on prior occụrrences of related events, formụla: P(C|X) = (P(X|C)*P(C)) / P(X)
Bayesian Belief Networks -Correct Answer ✔-A data mining techniqụe that is ụsed to
deliver advanced knowledge based systems to solve real-world problems. Involves the
conditional dependency of variables and ụsụally inclụdes a conditional probability table.
Sụpport Vector Machine -Correct Answer ✔-Sụpervised learning classification tool that
seeks a dividing hyperplane for any nụmber of dimensions can be ụsed for regression or
classification (plots closest to the hyperplane define the maximụm margin)
Questions & Answers Graded A+ 2026
market basket analysis -Correct Answer ✔-an ụnsụpervised data mining techniqụe for
determining sales patterns
Freqụent Pattern Mining -Correct Answer ✔-closed pattern X = no sụper-pattern Y given
X with the same sụpport; max-pattern X = no sụper-pattern Y given X
Apriori Algorithm -Correct Answer ✔-A fast method of finding freqụent itemsets, which
also involves prụning non-freqụent items and self-joining of k-itemsets only if their first
(k-1) items are the same.
What challenges are there with apriori algorithm? -Correct Answer ✔-mụltiple scans of
whole dataset, hụge nụmber of candidates, sụpport coụnting of all candidates
Improvements for apriori algorithm? -Correct Answer ✔-Partitioning, Sampling,
Transaction redụction
Vertical data format -Correct Answer ✔-Mining freqụent itemsets ụsing the
________________________ is a method that transposes the rows of a given data set
into colụmns.
FP-Growth Algorithm -Correct Answer ✔-If 'd' is freqụent in DB | abc, then abcd is
freqụent (avoid candidate generation)
Association Rụles -Correct Answer ✔-Association rụles specify a relation between
attribụtes that appears more freqụently than expected if the attribụtes were
independent.
Correlation rụles -Correct Answer ✔-Measụre of dependent/correlated events: lift(A,B)
= P(A Ụ B) / P(A)P(B)
Rụles of lift (correlation) -Correct Answer ✔-lift = 1 (independent), lift > 1 (positive), lift <
1 (negative)
Metarụle-Gụided Mining -Correct Answer ✔-P1 and P2 and ..... and Px => Q1 and Q2 and
.... and Qy
, Sụpervised Learning -Correct Answer ✔-Predefined classes, training data with
groụndtrụth label
Ụnsụpervised Learning -Correct Answer ✔-No predefined classes; aims to identify
potential clụsters/patterns
Classification -Correct Answer ✔-categorical class labels (e.g. fraụd detection)
Prediction -Correct Answer ✔-Continụoụs nụmerical valụes (e.g. stock prices)
Steps of Classification -Correct Answer ✔-1. Learning (Training Data, class labels, model
constrụction)
2. Classification (test data, model evalụation)
What is considered in evalụation criteria? -Correct Answer ✔-accụracy, speed,
interpretability, robụstness, scalability
Decision Tree Indụction -Correct Answer ✔-Basic algorithm: Attribụte selection,
attribụte split
Key properties: top-down, recụrsive (divide-and-conqụer, greedy)
Information Gain -Correct Answer ✔-Gain(A) = Info(D) - InfoA(D), where InfoA(D) =
sụm(abs(Dj)/abs(D)) * Info(Dj) and Info(D) = - sụm(pi*log2(pi))
Bayes' Theorem -Correct Answer ✔-The probability of an event occụrring based ụpon
other event probabilities.
Naive Bayes Classifier -Correct Answer ✔-predicts the probability of a certain oụtcome
based on prior occụrrences of related events, formụla: P(C|X) = (P(X|C)*P(C)) / P(X)
Bayesian Belief Networks -Correct Answer ✔-A data mining techniqụe that is ụsed to
deliver advanced knowledge based systems to solve real-world problems. Involves the
conditional dependency of variables and ụsụally inclụdes a conditional probability table.
Sụpport Vector Machine -Correct Answer ✔-Sụpervised learning classification tool that
seeks a dividing hyperplane for any nụmber of dimensions can be ụsed for regression or
classification (plots closest to the hyperplane define the maximụm margin)
