I need to help me to solve and understand this assignment .the requirements are :
1. Take theLoan_Original.ARFF
Download Loan_Original.ARFFfile describing the loan data. (I would like all of you to use the same original file, since I presume you have given different names of the attributes and it will be more difficult for me to comment).
For this assignment you will needtwo versionsof the loan data - the original file with numeric and nominal attributes, and a discretized one with nominal only attributes. Create different discretized versions as you did in Module Week 2, but using the new file,Loan_Original.ARFF
Download Loan_Original.ARFF. Use the discretized file that you think best represents the loan data (How about the one with "ALL" values? Note that in most cases when discretizing numeric attributes we lose information about the original data). ( I need a PrintScreen or screens for
all the works in Weka and explain the steps for this point ))
2. Using the Preprocess mode (the selected attribute bar graphs)find the best attribute(among both nominal and numeric). Use the approach outlined in
Handout Week 3
Download Handout Week 3“Notes on visualizing and analyzing instance spaces”. ( I need a PrintScreen or screens forall the works in Weka and explain the steps for this point )
3. Using the Visualize Panelfind the best pair of nominal attributesforeach of the two loan data sets(the original with numeric and nominal attributes, and the discretized one with nominal only attributes). Use the classification accuracy as an evaluation criterion and try several combinations of attributes (you may try all possible pairs or use the results from the single attribute selection for insights). For selecting attributes and computing accuracy use the approach described in
Handout Week 3
Download Handout Week 3“Notes on visualizing and analyzing instance spaces”. ( I need a PrintScreen or screens for
all the works in Weka and explain about the steps for this point ).
4. Write a short report describingthe approaches you used to find the best single attribute and the best pair of attributes for each data set and includethe two graphs(one for each set) produced by the Weka Visualization Panel and the corresponding accuracies. Include also any comment you may have on the results you obtained.