r_code

ch1

这个人写的书不错 * csgillespie/efficientR: Efficient R programming: a book 查看他的GitHub https://github.com/csgillespie

you need to know how long it takes to run. introduces the idea of benchmarking your code.

R optimizing for writing
C optimizing for running

因此说 C 比 R 快就弃用 R，是不科学的，我们需要比较 total time。简单来说，比较 C 和 R 导入 csv 文档的步骤。

library(tidyverse)

## -- Attaching packages --------------------------------------------- tidyverse 1.2.1 --

## √ ggplot2 3.1.0     √ purrr   0.2.5
## √ tibble  1.4.2     √ dplyr   0.7.8
## √ tidyr   0.8.2     √ stringr 1.3.1
## √ readr   1.2.1     √ forcats 0.3.0

## -- Conflicts ------------------------------------------------ tidyverse_conflicts() --
## x dplyr::filter() masks stats::filter()
## x dplyr::lag()    masks stats::lag()

library(data.table)

## 
## Attaching package: 'data.table'

## The following objects are masked from 'package:dplyr':
## 
##     between, first, last

## The following object is masked from 'package:purrr':
## 
##     transpose

knitr::include_graphics('pic/R_vs_C.png')

systime.time

system.time的定义

user time is the CPU time charged for the execution of user instructions.
system time is the CPU time charged for execution by the system on behalf of the calling process.
(\text{user} + \text{system} = \text{elapsed}) user 和 system 怎么理解？

`<-`和`=`的区别

Argument passing 和 Object assignment

<-都含有
=只有其一，因此容易报错

colon <- function(n) 1:n
system.time(res <- colon(1e8))

##    user  system elapsed 
##       0       0       0

system.time(expr = res <- colon(1e8))

##    user  system elapsed 
##       0       0       0

# system.time(res = colon(1e8))
# Error in system.time(res = colon(1e+08)) : 参数没有用(res = colon(1e+08))

system.time(expr, gcFirst = TRUE)里面没有参数res
1. expr - R Expression, 因此用于评价R表达代码的。
system.time(res <- colon(1e8)) = system.time(expr = res <- colon(1e8))

microbenchmark

microbenchmark 包参数和函数作用。避免自己手写 for loop 和 system.time

colon <- function(n) 1:n
seq_default <- function(n) seq(1, n)
seq_by <- function(n) seq(1, n, by = 1)
library(microbenchmark)
n <- 1e5
microbenchmark(
    colon(n)
    ,seq_default(n)
    ,seq_by(n)
    ,times = 10
)

## Unit: nanoseconds
##            expr     min      lq      mean    median      uq     max neval
##        colon(n)       0       0  130499.7    1026.5    3692 1289816    10
##  seq_default(n)    6974    8616  213451.5   20717.5  115690 1497811    10
##       seq_by(n) 1163049 1285303 1582362.5 1474016.5 1671756 2751524    10

read.csv vs readRDS

start_time <- Sys.time()
compare <- 
microbenchmark(
    read.csv('data/movies.csv')
    ,read_csv('data/movies.csv')
    ,fread('data/movies.csv')
    ,readRDS('data/movies.rds')
    ,read_rds('data/movies.rds')
    ,times = 10
) %>% 
    write_excel_csv('output/read_function_time.csv')
end_time <- Sys.time()
end_time - start_time
# Time difference of 10.29 secs

read_csv('output/read_function_time.csv') %>% 
    ggplot(aes(expr,time/1000000000)) +
    # because the microbenchmark unit is nano-second
    # 1000000000 nanosecond = 1 second
    geom_boxplot() +
    theme_bw() +
    theme(axis.text.x = element_text(angle = 70, hjust = 1)) +
    labs(
        caption =  "Made by Jiaxiang Li\nRound data with 10 times"
        ,x = "different read function"
        ,y = "time cost (s)"
        ,title = "fread is best!"
    )

## Parsed with column specification:
## cols(
##   expr = col_character(),
##   time = col_double()
## )

推荐使用fread

测试电脑

正如之前作者所说更新最新版本的R，会让代码运行更快。电脑的配置也会影响代码运行速度。

函数运行测试

library(benchmarkme)

res <- benchmark_std(runs = 3)
plot(res)
upload_results(res)

# PC laptop
res %>% 
    write_excel_csv('output/test_work_laptop.csv')
# PC desktop
res %>% 
    write_excel_csv('output/test_work_desktop.csv')

这里面都是一些数据集，进行电脑的测试。
封装了system.time函数

fread('output/test_work_laptop.csv') %>% 
    select(1:3) %>% 
    summarise_all(funs(sum(.)))

##    user system elapsed
## 1 114.1   0.87  115.84

fread('output/test_work_desktop.csv') %>% 
    select(1:3) %>% 
    summarise_all(funs(sum(.)))

##    user system elapsed
## 1 95.96   1.11   97.51

显然研发机会更快。

电脑配置

# Load the benchmarkme package
library(benchmarkme)

# Assign the variable ram to the amount of RAM on this machine
get_ram()

## 17.2 GB

# Assign the variable cpu to the cpu specs
get_cpu()

## $vendor_id
## [1] "GenuineIntel"
## 
## $model_name
## [1] "Intel(R) Core(TM) i5-6300U CPU @ 2.40GHz"
## 
## $no_of_cores
## [1] 4

读写能力

# Load the package
library(benchmarkme)

# Run the io benchmark
res <- benchmark_io(runs = 1, size = 5)

# Plot the results
plot(res)

不需要尝试>50MB，因为电脑测试不了。

res %>% 
    write_excel_csv('output/test_work_laptop_wr.csv')

fread('output/test_work_laptop_wr.csv')

##    user system elapsed   test test_group cores
## 1: 0.89   0.00    0.89  read5      read5     0
## 2: 1.90   0.03    1.95 write5     write5     0