COSC 667
Homework Problem


Question: can you predict if someone makes more than $50,000 per year? Use the data set here.

Please note: when you click on "Data Folder" you should then use the file adult.data.

What to do:
- give a prediction based upon Majority Rule
- give a prediction based on 1-Rule
- use ID3/J48 (weka) to build a tree


You will need to preprocess the data; there are missing data. You may also need to process the data to put it in arff format or a spreadsheet.

Question: can you trim the ID3/J48 tree and get results better than 1-Rule?

You should prepare a 2-4 page report on your findings. Please included any description of pre-processing. Also include a table clearly demonstrating the results from your algorithms. Describe the parameters you employed for J48 and describe all attempts to trim the tree. Do any of the algorithms yield results that you are happy with?