1. For the given loan data how k-means cluster (two clusters in Classes to clusters evaluation mode) how to compare the cluster centroids with the data models generated by the Weka classifiers, that is, if we consider these centroids as instances how are they classified according to the models produced by OneR or J48? 2. By using EM, Generate two clusters in Classes to clusters evaluation mode. Compare the clusters and their parameters with the class descriptions produced by the Naive Bayes classifier. Are they similar? 3. Cobweb: Use training set cluster mode and ignore the class attribute (loan approval). Describe only the top level split (the successors of the root) in the clustering hierarchy. Compute the classes to clusters evaluation error for the top level clustering by hand (see clustering experiments). Weka cannot be used for this purpose because it computes the evaluation for the leaf clusters only.

Loan Dataset:

ID	Sex	Age	Married	Employed	Money	Pay	Months	Buy	LastEmployer	Area	Approved
1	F	18	No	Yes	20	2	15	PC	1	good	Yes
2	F	20	No	Yes	10	2	20	PC	2	good	Yes
3	F	25	Yes	No	5	4	12	PC	0	bad	No
4	F	40	Yes	Yes	5	7	12	PC	2	good	Yes
5	F	50	No	Yes	5	4	12	PC	25	bad	Yes
6	M	18	No	Yes	10	5	8	PC	1	good	Yes
7	M	22	No	Yes	10	3	8	PC	4	good	Yes
8	M	28	Yes	Yes	15	4	10	PC	5	good	Yes
9	M	40	Yes	Yes	20	2	20	PC	15	good	Yes
10	M	50	Yes	No	5	4	12	PC	0	good	No
11	F	18	No	Yes	50	8	20	Car	1	good	No
12	F	20	Yes	No	50	10	20	Car	2	good	No
13	F	25	No	Yes	50	5	20	Car	5	good	No
14	F	38	No	Yes	150	10	20	Car	15	good	Yes
15	F	50	Yes	Yes	50	15	20	Car	8	good	Yes
16	M	19	No	Yes	50	7	20	Car	2	good	No
17	M	21	Yes	Yes	150	3	20	Car	3	good	Yes
18	M	25	No	Yes	150	10	20	Car	2	good	Yes
19	M	38	Yes	Yes	100	10	20	Car	15	good	Yes
20	M	50	Yes	Yes	50	10	30	Car	2	good	No