Analysis result of PCA

maryamshirzadmaryamshirzad Member Posts: 11 Learner I
edited June 2020 in Help

I want to use PCA. 
But I can't analyze the results.
Because the column names are displayed as "PC_1". And I don't know which feature the "PC_1" belongs to.
please help me.

Best Answer

  • Options
    BalazsBaranyBalazsBarany Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert Posts: 955 Unicorn
    Solution Accepted

    this is how PCA works. It calculates new principal components (PC_1 to PC_n) from the original attributes. 
    If you need the original attributes in addition to the principal components, you can do the following:
    1. Have or generate a unique ID in your data
    2. Set Role of this ID attribute to id so PCA doesn't change it
    3. Multiply
    4. Execute the PCA on one of the branches
    5. Join the processing branches back based on the ID attribute

    If you mark an attribute as label (with Set Role), the PCA operator won't change it. This allows you to build a model on the PCA attributes.

    I hope this helps. Here's a PCA video in the RapidMiner Academy:

    Best regards,


  • Options
    maryamshirzadmaryamshirzad Member Posts: 11 Learner I
    One Another question;Isn't the label feature used in PCA?
Sign In or Register to comment.