Log-Rank and Wilcoxon Algorithm
Suppose that the number of groups is 2. Denote the ordered observed failure times by
. At time
, let
be a positive weight function, and let
and
be the size of the risk set and the number of events in the jth sample, respectively. Let
.
Log-Rank and Wilcoxon statistic :
- is 1.0 for log-rank test
- is for wilcoxon test
covariance for Log-Rank and Wilcoxon statistic :
- is 1.0 for log-rank test
- is for wilcoxon test
Computation of :
- Compute the 100 quantiles of the values, denoted by .
- For (0, .01, .02, ... 1.0)
- Let , where be the percentile of the values.
- Compute , j = 1, 2, ... 100, where _mad_ is the median absolute deviation from the median, multiplied with 1.4826.
- Compute cv() = .
- Choose = argmin[cv()].
- Finally compute = . is henceforth fixed at the value .