SPSS Statistics

 View Only

  • 1.  Cross-validation for General/Generalized linear model in SPSS

    Posted Mon January 18, 2021 07:53 AM

    I am trying to predict the stress due to COVID using several predictors like age , race, gender, financial status. We got data from a survey. Some of the predictors are categorical, some ordinal. Trying to use General linear model or even Generalized linear model. As a first step, I am randomly selecting 80% of the cases so that I can use the remaining 20% for cross-validation. However, I do not find a way to do that or could not implement the model based on 80% to the remaining 20% using either General or Generalized linear model. Please help.

    Thanks.

    Pritish Mondal
    Penn State College of Medicine, Hershey



    ------------------------------
    Pritish Mondal
    ------------------------------

    #SPSSStatistics


  • 2.  RE: Cross-validation for General/Generalized linear model in SPSS

    IBM Champion
    Posted Mon January 18, 2021 09:27 AM
    Export your 80% model as xml (see the Export tab) and then use the Scoring Wizard (on Utilities) to score the other 20%

    --



    Original Message


  • 3.  RE: Cross-validation for General/Generalized linear model in SPSS

    Posted Mon January 18, 2021 12:09 PM
      |   view attached
    Thanks a lot Jon. Appreciate your help. I see a way to export generalized linear model (80%) as xml, however but not General linear model. 

    Thanks,
    Pritish

    ------------------------------
    Pritish Mondal
    ------------------------------

    Attachment(s)

    tif
    screenshot.tif   207 KB 1 version
    Original Message


  • 4.  RE: Cross-validation for General/Generalized linear model in SPSS

    IBM Champion
    Posted Mon January 18, 2021 12:24 PM
    If you can't use GENLIN, you can construct the predicted values for the new dataset with a compute based on the GLM parameter estimates.

    --



    Original Message