I am working on research paper for diagnosis of cancer.

List of Known prognostic factors

  1. Age of patient
  2. Size of tumor
  3. Grade of tumor
  4. Lymphnode involvement

and list of Unknown factors which are to be assessed with prognosis by correlating with known prognostic factors. u1,u2 and u3

I have 128 patient records for all three unknown factor. I have 231 patient records for u1 data is there and have missing u2 and u3 data.

I wanted to know what statistical test and data visualization I can use PCA or Multiple anova to show the relationship between the known and unknown factors.

  • $\begingroup$ Please specify what you mean by "show the relationship between the known and unknown factors." $\endgroup$
    – rolando2
    Oct 23, 2014 at 0:17
  • $\begingroup$ @rolando2 : Each cancer is being classified(diagnosis) using these known factor, I am trying to prove that unknown factor (new scale) that can be included and improve the diagnosis of cancer $\endgroup$ Oct 23, 2014 at 5:26

1 Answer 1


Showing the relationship between known and unknown factors is distinct from demonstrating that new factors improve the model. PCA will how all factors (but just the factors) are related in a reduced-dimension space. You could then regress the major PCs against the response variable. Or do multiple regression. Note that regression type will depend on the type of response data, which was not specified (yes/no?). Random forests are another possibility; not the liveliest discussion but good to get started: Random forest - binary classification vs. regression?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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