3
$\begingroup$

I need your help with a Statistical Learning homework in R. I have to perform classification over this dataset: mammographic masses predicting Severity (0="not severe",1 = "severe) using these predictors:

  • Age (quantitative)
  • Margin (qualitative)
  • Shape (qualitative)

Everything is fine and understandable when I use logistic regression, but I don't know if it's possible to run QDA (or linear discriminant analysis either), since two of the variables are qualitative.

$\endgroup$
1
$\begingroup$

My answer would be no you cannot use the usual form of LDA or QDA if your data points are binary. The reason for this is the following:

  • LDA and QDA require you to estimate several parameters of a multivariate gaussian distribution: the mean(s), the covariance matrix
  • How would you fit a multivariate gaussian on data that can only be equal to 0 or 1 ?

In his lecture notes, Andrew Ng specifically states that the assumption behind LDA is that the X data is continuous

See here:

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.