So this is my assignment. I havn't used Weka before, but I'm learning and just want to make sure I'm doing it right. So what I did was, open up testdata1.arff in Weka. Then I went to the 'classify' tab, hit 'choose' and chose one of the 13 classification methods and then under 'test options' I selected 'use training set'. And then a summary came up in the 'classifier output'. The car.arff had a whole bunch of data in it, and the testdata1.arff only had the 5 instances listed below in it. Why would he give us the car.arff if we're not even soppuse to use it? And what does he mean by summarize the results?
Using the Weka software, with the training data provided in car.arff (in datasets-UCI), determine the classification of the following five cars.
low,low,4,more,big,high,?
high,vhigh,3,2,small,high,?
vhigh,med,4,2,med,med,?
med,low,3,more,big,med,?
med,low,2,4,big,med,?
In order to do this, use the testdata1.arff file which has the predicted class arbitrarily set to unacc. Use the following 13 classification methods. For each method, summarize the results (predictions etc as given by Weka). Finally, provide a conclusive summary as to which one (in your opinion) is the most accurate.
1. BayesNet (Bayes) 2. NaiveBayes (Bayes) 3. Logistic (functions) 4. MultiLayerPerceptron (functions) 5. IBk (lazy) 6. LWL (lazy) 7. Bagging (meta) 8. Stacking (meta) 9. InputMappedClassifier (misc) 10. DecisionTable (Rules) 11. ZeroR (Rules) 12. DecisionStump (Trees) 13. J48 (Trees) [link][5 comments]