Cluster quality evaluation using follow-up data

Share:

Description

Function to evaluate the overall quality of a given partition using follow-up data. A partition (clustering) is composed of non-overlapping clusters.

Usage

1
surv_measure(parti, surv.time, status, method = "BIC")

Arguments

parti

A partition to be evaluated.

surv.time

A numeric vector contains follow-up time of patient's in x

status

A binary vector contains survival status of patients in x, normally 0=alive, 1=dead.

method

Type of partition evaluation measures to use for assessing the relationship between follow-up and a partition. Default is BIC.

Details

This function fits a Cox model using follow-up data as response and cluster labels in the partition as covariate. The likelihood from the fitted model further used to calculate the modified AIC or BIC. See references for more details. Note that, for convenience in later usage, returned value is multiplied by -1 inside the function so that large value denotes good quality partition.

Value

A numeric value representing the quality of partition under consideration in terms of follow-up.

Author(s)

Askar Obulkasim

References

Liang,H. and Zou,G.H. (2008). "Improved AIC selection strategy for survival analysis", Comput Stat Anal., 52, 2538-2548.

Volinsky,T.C. and Raftery,A.E. (2000). "Baysian information criteria for censored survival models", Biometrics, 56, 256-262.

Obulkasim,A. et al., (2013). "Semi-supervised adaptive-height snipping of the Hierarchical Clustering tree", submitted.

See Also

measure

Examples

1
2
3
4
5
data(BullingerLeukemia)
attach(BullingerLeukemia)
cl <- HCsnipper(em[, 1:30], min = 5)
cl <- cl$partitions[cl$id, ]
result <- apply(cl, 1, function(x) surv_measure(x, surv.time[1:30], status[1:30]))