How to use coxplsDR method to predict new data?

[psalguerog]

Hello fbertran,

I'm Pedro Salguero and I'm glad to use your R package, but I have some doubts I want to solve. I have been used the plsRcox method and it works perfectly. Also, I can run the predict method in R because it accepts plsRcoxmodel objects as you programmed.

However, when I run the coxplsDR method, the result is a coxph object that requires the different PLS components to predict new data and my new data have the original covariables.

I can run it with the parameter allres=TRUE that includes extra results, but I'm not able to use the information to obtain an automatic or manual prediction.

The method returns (in a list) a PLS object I can not use with predict function to obtain the component values for new data. I tried to get the components manually, but I do not understand wich values of the PLS object I should use to get them and then how to run the cox model. Can you help me with this method and how to use it to predict new data?

If you need more information about my problem I can add it.

Edit: Hello again, I'm also working with the function coxDKsplsDR adn again, I'm not able to make it work with new predictions. I get a list with a coxph, spls and pls models that I can not use to predict new observations.

I get the next errors if I try to predict the components with the spls or pls object.

Error in UseMethod("predict") : no applicable method for 'predict' applied to an object of class "spls"

Error in UseMethod("predict") : no applicable method for 'predict' applied to an object of class "pls"

Thank you,
Pedro Salguero