You are on page 1of 2

Ain Shams University

Faculty of Engineering
Computer and Systems Engineering Department

CSE 412: Selected Topics in Computer Engineering


4th Year CSE – 2nd Semester 2017/2018
SHEET 3

1. Consider the following data set that consists of four attributes A, B, C, and D. This data set is to
be grouped into three clusters.
i. Use K-means method to find the Centroid of each cluster and the samples belonging to each
cluster assuming samples 1, 3, and 6 are the initial Centroids.
ii. Write an R script that clusters this data based on K-means method. (mention your
assumptions)

2. Compare k-means clustering, Apriori association rules algorithm, and linear regression from the
following perspectives:
• Limitations
• Context of using each of them
• Advantages
• Disadvantages

3. The shown table provides a set of transactions for a database system. Specific items are
searched in these database transactions; namely A, B, C, D, and E. The shown table shows a set
of six transactions for this database with one or more of the required items appear in them. Find
all frequent itemsets and calculate their corresponding support assuming the minimum support
is one transaction. (B.Sc. May 2015)
4. Consider a database, D , consisting of 9 transactions. Suppose min. support required is 2 and
minimum confidence required is 70%. Find Association rules.

ASSIGNMENT
(Due next week)
An insurance company tries to rank its drivers based on several attributes, such as if the driver had
accidents in the past, if he lives in a city, and if he is well educated. Based on these three attributes it
categorizes the driver in one of two ranks either green rank or red rand. Statistical data have been
collected from this company for 1000 driver and the following table shows the number of drivers
belonging to each category according to their attributes. Assume that the minimum support is 30% and
the minimum confidence required is 50%. Use Apriori method to find all the association rules that can
be generated from the given data set.

You might also like