UNIVERSITI MALAYA

EXAMINATION FOR THE DEGREE OF MASTER OF DATA SCIENCE

ACADEMIC SESSION 2020/2021 : SEMESTER I

WQD7005: Data Mining

14th January 2020 from 8.00 am to 15th January 2020 5.00 pm

INSTRUCTIONS TO CANDIDATES:

Answer ALL questions (50 marks).

(This exam contains 4 pages including the first title page)

 WQD7005

PART A (30 marks)

1) Define "Data Mining" in terms of Business Intelligence (keeping in mind the data
transformation from Online Transaction Process (OLTP) to Online Analytic Process
(OLAP)).
(5 marks)

2) Suppose that the data for analysis includes the attribute age. The age values for the data
tuples are (in increasing order) 13, 15, 16, 16, 19, 20, 20, 21, 22, 22, 25, 25, 25, 25, 30, 33,
33, 35, 35, 35, 35, 36, 40, 45, 46, 52, 70. (5 marks)

a) What is the mean of the data? (1 mark)

b) What is the median? (1 mark)
c) What is the mode of the data? (1 mark)
d) Use smoothing by bin means to smooth the above data, using a bin depth of 3.
Illustrate your steps. (2 marks)

3) Suppose you have the following four Dimension Tables namely Time, Customer, Employee
and Product. Construct a snowflake scheme by developing "Sales" Fact Table. The linkage
attribute in the dimension tables can be used to split the table to form a snowflake scheme.
The aggregate variable of fact table can be "quantity" of products.

Time Customer
OrderID (primary key) CustID (primary key)
Order Date Name
Year Address
Quarter CityID (linkage attribute)
Month City Name
Zip Code
State
Country

Employee Product
EmpID (primary key) ProductID
Employee Name Product Name
DepartmentID (linkage attribute) Product Category
Region Product Description
Territory

(5 marks)

4) Suppose you have the following transactional database, construct an FP (frequent pattern)
tree from this transaction database.
(5 marks)

2/4
 WQD7005

5) Let us consider the dataset of sales related to computer systems (e.g. hardware and software)
shown below. We are required to learn a decision tree which predicts the profit either up or
down based on certain features i.e. condition, upgradable and type.
(5 marks)

Condition Upgradable Type Profit

Old Yes S/W Down
Old No S/W Down
Old No H/W Down
Mid Yes S/W Down
Mid Yes H/W Down
Mid No H/W Up
Mid No S/W Up
New Yes S/W Up
New No H/W Up
New No S/W Up

Calculate the Information Gain of feature "Condition" based on,

Entropy (Profit)
Entropy (Old)
Entropy (Mid)
Entropy (New)
Entropy (Condition)

6) Write down the steps of DBScan algorithm.

(5 marks)

3/4
 WQD7005

PART B (20 marks)

Instructions: Answer the following questions by using any data mining tool. Explain how you
do each step (include print screens). Download “Data(Exam).csv” from the Spectrum (You can
find the description of this data at https://archive.ics.uci.edu/ml/datasets/Zoo).

1) Select the best non-target features using one of statistical methods "correlation", "Chi-
square", or "ANOVA". Your solution should describe the relevant statistical findings.
(5 marks)

2) Experiment/simulate the classification algorithms (Naive Bayes, Random Forest, Support

Vector Machine) and identify the best algorithm among the three algorithms using 10-fold
cross validation. Justify your choice of algorithm in terms of classification accuracy and
false positive rate.
(10 marks)

3) Discuss the performance metric of all three algorithms in terms of Receiver Operator
Characteristic (ROC) curve.
(5 marks)

END

4/4