I' m doing a survival analysis using machine learning with random forest method.
There are two groups for comparison.
But two groups are so unmatched in baseline characteristics. I performed propensity score matching with 1:1 ratio which eliminated all inequalities(assess by ASMD plot). But my sample volume dropped to 30%.
2:1 ratio PSM was not satisfactory regarding to matching performance.
Is it necessary to match data groups before random forest?

  • $\begingroup$ Can you show some sample data from each group? $\endgroup$ Apr 1, 2021 at 19:12


Your Answer

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