WEKA job assignment 2 -- 2

Extract principle components of the training dataset given using the “Principal Component Analysis(PCA)” feature in WEKA tool. How to use PCA in WEKA: 1. Load the training dataset file from the “preprocess” tab. 2. Go to “Select attributes” tab, from “Attribute Evaluator” choose “PrincipalComponents”. 3. Upon selection of PrincipalComponents, a popup window will open to change the search method to “Ranker” (If it was not already selected). Click “yes”. 4. Click on the “start” button. Upon on doing this the WEKA tool will generate the principal components of the dataset given. 5. Save the Transformed dataset. Now you need to calculate the Principal components of the “test dataset”. In order to apply any data mining algorithm on training and test datasets, both the datasets must have similar attributes. Unfortunately using Weka GUI tool we can only transform either training or test dataset but not both in to the same dimension. So we need to run weka command called batch filtering command to do that. You can find instructions about it in the following link https://weka.wikispaces.com/Batch+filtering The command that you see in the above link is for "Standardize" filter, but you need to modify the command to use "PrincipalComponents" filter. Below is the modified command java [login to view URL] -b -i [login to view URL] -o [login to view URL] -r [login to view URL] -s [login to view URL] After successful execution of the command, you will get transformed training and test datasets. You can use these datasets and apply the data mining algorithms like you did in your assignment-2 (this is just for your info, you no need to do this for your assignment). In your answer mention the number of attributes in the original test and training data and the number of attributes in the transformed test and training data (obtained using the batch filtering). Also submit the transformed test and training dataset files that you obtained by executing the batch filtering command.