CIND110 Assignment3-Association rules Solved

Description

1. Association rules:

One of the major techniques in data mining involves the discovery of association rules. These rules correlate the presence of a set of items with another range of values for another set of variables. The database in this context is regarded as a collection of transactions, each involving a set of items, as shown below.

Trans ID      Items Purchased

• Meat, Potato, Onion
• Meat, Noodle
• Noodle, Spinach
• Meat, Potato, Onion
• Onion, Potato, Noodle
• Eggs, Spinach
• Eggs, Noodle
• Meat, Potato, Salt, Onion
• Salt, Spinach
• Meat, Potato
• Apply the Apriori algorithm on this dataset.

Note that, the set of items is {Meat, Potato, Onion, Noodle, Spinach, Eggs, Salt}.  You may use 0.3 for the minimum support value.

• Show the rules that have a confidence of 0.8 or greater for an itemset containing three items.

2. Classification:

Classification is the process of learning a model that describes different classes of data and

the classes should be pre-determined. Consider the following set of data records:

ID	Age	City	Gender	Education	Profile
101	20-30	NY	F	College	Employed
102	31-40	NY	F	College	Employed
103	51-60	NY	F	College	Unemployed
104	20-30	LA	M	High School	Unemployed
105	41-50	NY	F	College	Employed
106	41-50	NY	F	Graduate	Employed
107	20-30	LA	M	College	Employed
108	20-30	NY	F	High School	Unemployed
109	20-30	NY	F	College	Employed
110         51-60          SF             M        College                       Unemployed

 

Assuming, that the class attribute is Profile, apply a classification algorithm to this dataset.

3. Clustering: Consider the following set of two-dimensional records:

 

RID                  Age                    Years of Service

• 30 5
• 50 25
• 50 15
• 25 5
• 30 10
• 55 25

 

• Marks:

Use the K-means algorithm to cluster this dataset. You can use a value of 2 for K and can assume that the records with RIDs 103, and 104 are used for the initial cluster centroids.

• Marks:

What is the difference between describing discovered knowledge using clustering and describing it using classification?