CSC 470 Data Mining Fall 2005 – Evening Section

Assignment 1                            Data Preparation                                  10 points

Assigned: 09/14/05

Due:         09/21/05 

Task:

            Create arff files suitable for the WEKA software as described below:

1. Obtain the file NJDOHcleanFinalDataReducedReady.csv from my WWW page

2. Remove attribute: FIPScode

3. Use Excel to replace 1990 data (5 attributes) with percent change from 1990-2000 (actually use a decimal such as .02 would be a two percent increase)

· Be careful!  If you don’t know how to do this relatively quickly in Excel – ask!

· Remove DIV / 0 errors – replacing with ? for unknown

4. Put 2000 drug admissions last (except for % town name)

5. Save the adjusted file as a csv file, but with a arff extension

6. Edit the arff file using Wordpad to add arff header information. Make Drug Admissions a Nominal with possible values 0,1,2, etc

7. Replace empty data (how does Excel save empty data when saving to csv?) with ? for unknown.

8. Save, of course

9. Open your file in WEKA to ensure that it opens correctly

10. Discretize all numeric attributes using WEKA –

a. Use Unsupervised Attribute Discretize under the Preprocess tab

b. Choose options bins=5 and FindNumBins=True

11. Look at visualization in Preprocess area to ensure discretizing worked

12. Save to a different file name so that you have both versions of the file

13. Re-open the discretized file to make sure that that opens.

14. Turn in both files you created on a disk