Module1: Classification

Use the IRIS dataset that comes with Weka to compare classification performance of the following algorithms.

•Naive Bayes Simple

•Multi Layer Perceptron

•J48 (C4.5, decision tree induction)

For each classifier, run test once using the training set for testing, and once using 5-fold cross-validation.

•prepare a short report which describes your work. This report must address the following issues.

i) Description of the dataset.

ii) A table describing the results.

iii) A discussion/explanation of the results and the algorithms’ performance. Also, describes why the accuracy is higher when “use training set” is selected for testing, rather than “cross-validation”.

Module 2: Clustering

Compare three clustering algorithms in Weka.

For this comparison, you would require to use at least two different datasets.

Run the algorithms on datasets, and use the visualization tools.

•prepare a short report which describes your work. This report must address the following issues.

i) Description of the dataset.

ii) Description of the algorithms used.

iii) Discussion of the clustering approaches, and the information that you might gain from doing this clustering.

iv) Summary of the results.

v) Comparison between the algorithms used, in terms of what you think is interesting.