Lecture 3

Question 1

What are the 5 steps of the data preparation process?

Answer 1

Collect data Prepare data Build model Evaluate model Deploy model

Question 2

Name 7 data preprocessing techniques

Answer 2

Feature subset selection Discretization and binarization Dimensionality reduction Aggregation Attribute creation and transformation Sampling

Question 3

Explain the preprocessing technique, aggregation.

Answer 3

Combining two or more attributes (or objects) into a single attribute (or object).

Question 4

What are two techniques of aggregation?

Answer 4

Data reduction: reduce the number of attributes or objects Change of scale: cities are aggregated into countries, etc…

Question 5

Describe the 4 sampling techniques used in sampling for preprocessing data?

Answer 5

Simple Random Sampling - each object is selected with equal probability) Sampling without Replacement - remove object from sample if selected Sampling with Replacement - do not remove object from sample if selected Stratified Sampling - split the data into several partitions and take random samples from each partition

Question 6

What are the two techniques of feature subset selection?

Answer 6

Remove redundant features - duplicate much or all of the information contained in one or more other attributes eg. purchase price of a product and the amout of sales text paix Remove irrelevant features - contain no informatino that is useful for the datamining task at hand eg. student id’s is often irrelevant to the task of predicting students’ GPAs

Question 7

What are the 3 techniques of discretization without using class labels?

Answer 7

Equal interval width Equal frequency K-means

Question 8

Describe the curse of dimensionality

Answer 8

When the dimensionality increases, data becomes increasingly sparse in the space that it occupies

Question 9

What is Big O’ notation?

Answer 9

Measure to evaluate the performance of an algorithm with respect to space or time Used to describe how an algorithm behaves with respect to run time and/or space as the problem size grows.

Question 10

What is the name of this type of Big O notation? O ( 1 )

Answer 10

Constant

• something can be done in constant time
  eg. pick a number from a data set and return it

Question 11

What is the name of this type of Big O notation? O ( n )

Answer 11

Linear eg. Finding an item in an unsorted list

Question 12

What is the name of this type of Big O notation? O ( n log n )

Answer 12

Linearthimic

eg. performing a mergesort or heapsort

and used by decision trees

Question 13

What is the name of this type of Big O notation?

O (n²)

Answer 13

Quadratic

Question 14

What is the name of this type of Big O notation?

O (n³)

Answer 14

Cubic

Question 15

What is the name of this type of Big O notation?

O ( 2ⁿ)

Answer 15

Exponential

eg. used for feature subset selection

Question 16

What is the name of this type of Big O notation?

O ( n! )

Answer 16

Factorial

• used for permutation based algorithms