Due to recent updates, all users are required to create an Altair One account to login to the RapidMiner community. Click the Register button to create your account using the same email that you have previously used to login to the RapidMiner community. This will ensure that any previously created content will be synced to your Altair One account. Once you login, you will be asked to provide a username that identifies you to other Community users. Email us at Community with questions.
Selecting Bottom Principle Components
Hi Guys,
I am interested in using the bottom principle components (small non-zero eigenvalues) for outlier analysis but can't seem to
find a way to extract them either from the PCA model or from the PCA operator.
Is the pre-processing model generated by PCA the same regardless of the parameters selected, in other words are all principle components computed and then only a subset reported?
Also, is there a way to see the actual eigenvalue not just the rank PC1, PC2, ... :P
Thanks,
-Gagi
I am interested in using the bottom principle components (small non-zero eigenvalues) for outlier analysis but can't seem to
find a way to extract them either from the PCA model or from the PCA operator.
Is the pre-processing model generated by PCA the same regardless of the parameters selected, in other words are all principle components computed and then only a subset reported?
Also, is there a way to see the actual eigenvalue not just the rank PC1, PC2, ... :P
Thanks,
-Gagi
0
Answers
this is currently not possible but you might use the scripting operator to retrieve the values...
Greetings,
Sebastian
Thanks for the reply. I was also wondering if the PCA model is the same regardless of the parameters used to generate it.
Basically does RM compute the entire set of PCs for with no dimensionality reduction regardless of the dimension reduction selected in the PCA operator.
In terms of scripting, I believe you are talking about groovy scripts which I have to take a look at.
Thanks,
-Gagi
I have managed to select the bottom relevant (non zero variance) principle components by preforming PCA with no dimensionality reduction and using the select attributes operator with regular expressions. ex: pc_[1-9] would select the first 9 principle components. You may also want to look at the pre-processing model to see what variance each component contributes.
I still cant extract the eigenvalues for each PC but I think that will require some groovy script to access the RM data structure. ???
-Gagi
this is indeed the case.
Greetings,
Sebastian