Key Words

Question 1

Data Mining

Answer 1

identifying implicit, previously unknown, potentially useful patterns in data

Question 2

Data Mining can be done…

Data Mining can be…

Answer 2

Data Mining can be done… interactively or automatically

Data Mining can be… descriptive or predictive

Question 3

Machine Learning (for data mining)

Answer 3

programs that induce structural descriptions from observations

Question 4

Supervised learning

Answer 4

is based on labeled examples and used for predicting labels of new observations

Question 5

Unsupervised learning

Answer 5

is based on unlabeled data

Question 6

Classification Rule

Answer 6

predicts value of given attribute

Question 7

Association Rule

Answer 7

predicts value of arbitrary attribute (or combination)

Question 8

Concepts

Answer 8

structures that can be learned/thing to be learned

Question 9

Concept Description

Answer 9

Output of learning scheme

Question 10

Important types of learning problems (four)

Answer 10

Classification Learning
Numeric Prediction
Association Learning
Clustering

Question 11

Clustering

Answer 11

Grouping similar instances into clusters

Question 12

Is classification learning supervised or unsupervised?

Answer 12

Supervised - Scheme is provided with actual outcome

Question 13

Explain classification learning

Answer 13

-predicting a nominal class
Can measure success on fresh data for which class labels are known (test data) - in practice success is often measured subjectively.

Question 14

What do we call the outcome?

Answer 14

The class of the example

Question 15

Regression/Numeric Prediction

Answer 15

predicting a numeric quantity

Variant of classification learning where “class” is numeric

Question 16

Is regression/numeric prediction supervised or unsupervised

Answer 16

supervised - scheme is being provided with target value

Question 17

Explain numeric prediction/regression

Answer 17

Meaure success by comparing predictions to actual values in test data

Question 18

Association Learning

Answer 18

Detecting associations between attributes
Can be applied if no class is specified and any kind of structure is considered "interesting"

Question 19

is association learning supervised or unsupervised

Answer 19

Unsupervised

Question 20

What is the difference between association learning and classification learning?

Answer 20

Association learning can predict any attributes value, not just the class, and more than one attributes value at a time
Consequence: far more association rules than classification rules

Question 21

Clustering

Answer 21

Finding groups of items that are similar

success often measured subjectively

Question 22

Is clustering supervised or unsupervised

Answer 22

unsupervised - the class of an example is not known

Question 23

Instance

Answer 23

example provided as a fixed-length row of data

• thing to be classified, associated or clustered
• individual, independent example of target concept
• Characterized by a predetermined set of attributes

Question 24

What is a dataset?

Answer 24

Input to learning scheme, set of instances

Question 25

Denormalization

Answer 25

process of flattening, several relations are joined together to make one, denormalization is possible with any finite set of finite relations and may produce spurious regularities that reflect structure of database

Question 26

Denormalization problems

Answer 26

• relationships without pre-specified number of objects

e. g concept of nuclear family

Question 27

Concept of nuclear family

Answer 27

may need relational data mining techniques, which are beyond the scope of this course