Data mining

Question 1

name qualitative

Answer 1

nominal
ordinal

Question 2

name quantitative

Answer 2

interval
ratio

Question 3

Data preprocessing

Answer 3

aggregation
sampling
dimensionality reduction
feature subset selection
feature creation
discretization and binarization
attribute transformation

Question 4

aggregation

Answer 4

combining two or more attributes

Question 5

types of sampling

Answer 5

simple random sampling
sampling with replacement
sampling without replacement
stratified sampling

Question 6

dimensionality reduction

Answer 6

PCA
singular value decomposition

Question 7

feature subset selection

Answer 7

brute-force approach
embedded approach
filter approach
wrapper approach

Question 8

attribute transformation

Answer 8

standardization
normalization

Question 9

pro MIN

Answer 9

can handle non-elliptical shapes

Question 10

limitation MIN

Answer 10

sensitive to noise and outliers

Question 11

pro MAX, group average, ward’s method

Answer 11

less susceptible to noise and outliers

Question 12

limitation average group, ward’s method

Answer 12

biased towards globular clusters

Question 13

limitation MAX

Answer 13

tends to break large clusters
biased towards globular clusters

Question 14

4 advantages of using decision tree

Answer 14

inexpensive to construct
extremely fast for classifying unknown records
easy to interpret
accuracy is comparable to others

Question 15

4 disadvantages of using decision tree

Answer 15

do not generalize well to certain boolean functions
the used induction algorithm is greedy
not expressive enough for modeling continuous variables
tree replication

Question 16

4 disadvantages using MAX

Answer 16

tendency to break large clusters
biased towards global clusters
once a decision is made, can’t be undone
no objective function is minimized

Question 17

classification techniques

Answer 17

decision tree
rule-based
memory-based
neutral networks
support vector machines