Hi,
I doing some analysis and truying to use Survival analysis (Survival
Package) to show the relation between some metrics and the presence of an
event A in a classe.
I build a Cox model with Coxph.
My problem is with the function predict.
Because i would like to know the accuracy of the model (by calculating for
example the precision and the recall).
I have some classes in several versions and i know in which version i have
an event A, so i would like to predict with my cox model the presence or not
of an event A in the classe and compare it to the reality to know that if
with the model i can predict correctly the existence or not of the event A.
I don't know if this is possible with the cox model with R. I try to use
predict(cox.eclipse, type="lp") but i get in my result some negative values
and i wanted to compute the correlation between this prediction and the
status value (which is boolean 0 = without event A, 1= with event A) for
each subject (this doesn't work because the correlation was very small 0.16)
Now i'm confused, i don't know how i can prove that my model is accurate
like for other machine learning technics by computing the number of false
positive, false negative, precision and recall.
Your help will be very appreciated.
Thank you very much.
Regards,
Abdou Maiga
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
