lapply(X,median) which computes the median of each variable in the data frame X (there’s no need for a margin argument). apply(X,2,median) to compute the median of the columns of X, you should consider using lapply instead because it’s much faster, e.g. If you’re using apply on a data frame, e.g. The family of apply functions ( apply, lapply, tapply, sapply, etc.) in R provide an extremely convenient way of applying a function to the margins of an array, matrix, list, etc. This note discusses how incorporate parallel and associated packages, with little or no additional effort on the part of the statistical practitioner, to speed up data processing and statistical analysis pipelines. The parallel package builds on multicore and snow to provide a (mostly) platform agnostic method of leveraging multiple cores to speed up the computation of embarrassingly parallel problems Since 2011 R has supported parallel computation as part of the base distribution with introduction of the parallel package (R version 2.14.0 released in October 2011). If you tried to parallelise your R code a few years ago, you probably worked with the architecture specific snow (Windows) or multicore (Unix-like) packages. Parallel computation in R has come a long way over the last 10 years. Parallel computing is particularly suitable for ‘single program, multiple data’ problems, for example in simulations and bootstrapping. Parallel computing refers to situations where calculations are carried out simultaneously, for example distributing the calculations across multiple cores of your computer’s processor, as opposed to having the calculations run sequentially on a single core. Note: This first appeared in Biometric Bulletin Volume 33, Issue 2, Software Corner.
0 kommentar(er)
