R & C++ integration

这题目太标题党了，最近啃了半部《C++ Primer Plus》，除了书中习题，就没其它写过C++程序，下午读了Rcpp的文档，把mlass包里kMeans算法的核心部分用c++重新写了。

C++果然给力，速度提高不是一点两点。

> require(mlass)
> data(iris)
> x=iris[,-5]
> res = benchmark(kMeans(x, 3, lang="CPP"), 
+ kMeans(x, 3, lang="R"), 
+ columns=c("test", "replications", "elapsed", 
+ "relative", "user.self", "sys.self"))
> res
                        test replications elapsed relative user.self sys.self
1 kMeans(x, 3, lang = "CPP")          100   0.176  1.00000     0.173    0.003
2   kMeans(x, 3, lang = "R")          100  11.343 64.44886    11.293    0.015

Related Posts

• December 26, 2011 -- 用kmeans算法压缩图片 (1)
• December 24, 2011 -- ML-Class Ex 7 – kMeans clustering (2)
• April 12, 2012 -- 支持向量机 (0)
• August 3, 2012 -- project euler — problem 54 (2)
• October 9, 2012 -- project euler – problem 60 (0)