Add logrank test as query (scoring column)

[mstreit]

For finding significant changes in survival between clusters http://en.wikipedia.org/wiki/Logrank_test

[ngehlenborg]

For testing purposes we should implement this also for any column that uses the Kaplan-Meier plots.

[ngehlenborg]

Is there a Java implementation somewhere? With the apparent trouble that R integration is causing at the moment that would be a better solution.

[sgratzl]

basic implementation in r5677, r5678 (ui), r5683

based on: http://books.google.at/books?id=7Vg5wTSdwuMC&printsec=frontcover&dq=Survival+Analysis:+A+Self-Learning+Text&hl=de

[sgratzl]

@nils, marc: can anyone test this score, if it makes sense?

[sgratzl]

to sum up: log rank in this context is a metric of a group, defined as the logrank between the elements in the group versus all the others elements in the same stratification, where both of course are mapped to the clinical data domain including removing invalid samples

@marc: wan't more

[sgratzl]

additional column needed: p-value of the log rank test

see http://stat.ethz.ch/education/semesters/ss2011/seminar/contents/presentation_2.pdf and book again

e.g. upon creation of a log-rank column create three columns at once:

• the log rank value
• the p-value of the log rank test
• a combined one with: log-rank * p-value

[sgratzl]

r5703

according to: Survival Analysis: A Self Learning Text: {{{ The null hypothesis being tested is that there is no overall difference between the two survival curves. Under this null hypothesis, the log–rank statistic is approximately chi-square with one degree of freedom. Thus, a P-value for the log–rank test is determined from tables of the chi-square distribution. }}}

using the http://weka.sourceforge.net/doc/weka/core/Statistics.html from weka