| compute.external {StabPerf} | R Documentation |
Either sampling features from a selector or errors from a feature-selection/classifier-combination
compute.external(parameters)
parameters |
the job (a list of parameters) that should be computed with the elements offset, repeats, output, i (number of cv-fold), s (number of selector), selector, c (number of classifier), classifier, predictor |
Reads in the dataset from rawdata or the filesystem. If i > 0, cv-information is loaded from cvdata or the filesystem and afterwards selecting only the data not in the testings subset of the i-th fold. Then either features are sampled from a draw subset of the data (given a selector-job) or classifiers are trained and tested, using the previously sampled features/drawn indices (selector/classifier-job). Both is done repeats times, finally writing the sampled results to the corresponding checkpoint.
Invisible |
.compute.jobs, pipeline, .select, .classify.and.predict
lappyl(.job.queue(...), compute.external)