iSixSigma

Multiple regression

This topic contains 3 replies, has 4 voices, and was last updated by  Hody 9 years, 5 months ago.

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
    Posts
  • #53432

    Gadgil
    Member

    In a scenario where we have Y as a discrete type of data(no of defects) & Xs are a combination of continuous type of data & ordinal data,can we use mutiple regression equation to establish a relationship between Y & Xs?
    If multiple regressio can’t be used what approach should be taken?

    0
    #190069

    bbusa
    Participant

    Try Logistics Regression –

    pl clarify what exactly the Xs are . You may need to define dummy variables

    0
    #190073

    Robert Butler
    Participant

    If the Y is actually a defect count then standard multiple regression methods will suffice. As for the X’s if the values are ordinal, as opposed to nominal, then you can treat them as you would any other continuous X with the proviso that they can only take on the values used in the analysis. If the X’s are nominal then, as noted previously, you will have to build dummy variables and use these in place of the nominal values for the X variable of interest.

    You will want to pay close attention to the results of your residual analysis because this is the kind of a situation where there is a good chance the residuals won’t be normally distributed which will mean you will likely have to examine possible transforms of your Y or some of your X’s or perhaps both and build a final model using transformed values.

    0
    #190075

    Hody
    Member

    In Minitab, General Linear Modeling (GLM) can probably be used as an alternative since it can handle ordinal Xs (as Factors), and continuous Xs (as Covariates). Also the GLM will determine the statistical significances between the Xs…

    Have fun!

    0
Viewing 4 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic.