关于apply家住大家总是很困惑，我将自己的理解写出来，有更好的建议，欢迎留言。

apply家族

apply(array, margin, …)输入array输出vetor

margin:1表示行， 2表示列

  1. mat = matrix(1:24,nrow = 4,ncol = 6);mat
  2. ##      [,1] [,2] [,3] [,4] [,5] [,6]
  3. ## [1,]    1    5    9   13   17   21
  4. ## [2,]    2    6   10   14   18   22
  5. ## [3,]    3    7   11   15   19   23
  6. ## [4,]    4    8   12   16   20   24
  7. apply(mat, 1, sum)
  8. ## [1] 66 72 78 84

lapply(list, function) 输入list遍历list的元素，返回的也是list。

适合回归函数lm，因为lm返回也是list。

  1. x = list(1:5)
  2. lapply(x , FUN = log) 
  4. #output
  5. [[1]]
  6. [1] 0.0000000 0.6931472 1.0986123 1.3862944 1.6094379
  8. colnames(iris)
  10. # output
  11. [1] "Sepal.Length" "Sepal.Width"  "Petal.Length" "Petal.Width" 
  12. [5] "Species"   
  14. lapply(iris[,1:3], function(x) lm(x~iris$Sepal.Width, data = iris))
  16. # output
  17. $Sepal.Length
  19. Call:
  20. lm(formula = x ~ iris$Sepal.Width, data = iris)
  22. Coefficients:
  23.      (Intercept)  iris$Sepal.Width  
  24.           6.5262           -0.2234  
  27. $Sepal.Width
  29. Call:
  30. lm(formula = x ~ iris$Sepal.Width, data = iris)
  32. Coefficients:
  33.      (Intercept)  iris$Sepal.Width  
  34.        2.321e-15         1.000e+00  
  37. $Petal.Length
  39. Call:
  40. lm(formula = x ~ iris$Sepal.Width, data = iris)
  42. Coefficients:
  43.      (Intercept)  iris$Sepal.Width  
  44.            9.063            -1.735

sapply(list, function) 输入list，返回向量和矩阵，数据框

  1. sapply(1:5,log)
  2. #output
  3. [1] 0.0000000 0.6931472 1.0986123 1.3862944 1.6094379       
  5. sapply(1:5, function(x)(x+3))       
  6.  #output
  7. [1] 4 5 6 7 8

tapply(vector, index, function) 输入vector，输出array

index表示按什么切分，将一个连续变量根据一个分类变量进行切分，并分类汇总

  1. tapply(X = iris$Sepal.Length,INDEX = iris$Species,FUN = mean) #不能使用小写x
  2. #output
  3. setosa versicolor  virginica 
  4. 5.006      5.936      6.588

mapply(function, …) 输入函数，输出vector

操作函数，mapply具有向量化操作的功能

  1. myfun = function(x,y) {
  2.   if(x>4) return(y)
  3.   else return(x+y)
  4. }
  5. myfun(1:5,2:6) 直接运行会报错，因为此函数没有向量化操作的功能
  6. #output
  7. [1]  3  5  7  9 11
  8. Warning message:
  9. In if (x > 4) return(y) else return(x + y) :
  10.   条件的长度大于一，因此只能用其第一元素
  12. mapply(myfun, 1:5,2:6)
  13. #output
  14. [1] 3 5 7 9 6

tips-从data.frame里以data.frame取列和以vector取列

  1. > head(iris$Sepal.Length)
  2. [1] 5.1 4.9 4.7 4.6 5.0 5.4
  4. > class(head(iris$Sepal.Length))
  5. [1] "numeric"
  7. > class(head(iris["Sepal.Length"]))
  8. [1] "data.frame"
  10. > head(iris[,1])
  11. [1] 5.1 4.9 4.7 4.6 5.0 5.4
  13. > head(iris["Sepal.Length"]) #data.frame
  14.   Sepal.Length
  15. 1          5.1
  16. 2          4.9
  17. 3          4.7
  18. 4          4.6
  19. 5          5.0
  20. 6          5.4
  22. > class(head(iris["Sepal.Length"]))
  23. [1] "data.frame"
  25. > head(iris[,1:2]) #去两列那自然就是data.frame
  26.   Sepal.Length Sepal.Width
  27. 1          5.1         3.5
  28. 2          4.9         3.0
  29. 3          4.7         3.2
  30. 4          4.6         3.1
  31. 5          5.0         3.6
  32. 6          5.4         3.9
  34. > class(head(iris[,1:2]))
  35. [1] "data.frame"

dply包

以下函数的记忆法：第一个字母为输入数据结构，第二个字母为输出数据结构，a=array，d = dataframe，l=list
aaply(my_matrix,.margins = 2,.fun=mean) 输入array输出array
adply(my_matrix,.margins = 2,.fun=mean) 输入array输出data.frame
laply(my_list,.fun = mean)输入list 输出vector
ddply(.data = my_df,.variables =.(gender),.fun = mean) 输入dataframe，输出dataframe
dlply(iris,~Species,my_mode)适合回归分析,输入数据框，输出list

  1. library(plyr)
  2. my_matrix = matrix(1:24,nrow = 3,ncol = 8)
  4. > my_matrix
  5.      [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8]
  6. [1,]    1    4    7   10   13   16   19   22
  7. [2,]    2    5    8   11   14   17   20   23
  8. [3,]    3    6    9   12   15   18   21   24

aaply(my_matrix,.margins = 2,.fun=mean) #输入array输出array

  1. > aaply(my_matrix,.margins = 2,.fun=mean) #输入array输出array
  2.  1  2  3  4  5  6  7  8 
  3.  2  5  8 11 14 17 20 23 
  6. #和apply比较
  7. > apply(my_matrix, 2,mean) #输入array输出vector
  8. [1]  2  5  8 11 14 17 20 23

adply(my_matrix,.margins = 2,.fun=mean) #输入array输出data.frame

  1. > adply(my_matrix,.margins = 2,.fun=mean) #输入array输出data.frame
  2.   X1 V1
  3. 1  1  2
  4. 2  2  5
  5. 3  3  8
  6. 4  4 11
  7. 5  5 14
  8. 6  6 17
  9. 7  7 20
  10. 8  8 23

laply(my_list,.fun = mean)输入list 输出vector

  1. > my_list = list(1:10,2:8,rep(c(T,F),times = 5))
  2. > laply(my_list,.fun = mean)
  3. [1] 5.5 5.0 0.5
  5. #和lapply比较
  6. > lapply(my_list, mean) # 输入list输出array
  7. [[1]]
  8. [1] 5.5
  10. [[2]]
  11. [1] 5
  13. [[3]]
  14. [1] 0.5

ddply(.data = my_df,.variables =.(gender),.fun = mean) 输入dataframe，输出dataframe

  1. > my_df = data.frame (name= c('Tony','Andy','Bob', 'Mary', 'Leo'),
  2. + height=c(178,176,175,167,199),
  3. + gender= c('M','F','F','M','M'),
  4. + age = c("old","young","young","old","young"))
  6. > ddply(.data =  my_df$sheight,.variables =.(gender),.fun = mean) #报错，此处输入的是vector
  7. NULL
  9. > ddply(.data =  my_df,.variables =.(gender),summarise,
  10. +       mean_h= mean(height))  #.data 输入的是整个dataframe，.(gender)是此包的格式
  11.   gender   mean_h
  12. 1      F 175.5000
  13. 2      M 181.3333
  15. > ddply(.data =  my_df,.variables =.(gender),summarise,
  16. +       mean_h= mean(height),sd_h = sd(height)) #同时汇总mean 和sd
  17.   gender   mean_h       sd_h
  18. 1      F 175.5000  0.7071068
  19. 2      M 181.3333 16.2583312
  21. #和tapply比较，输入向量，输出array
  22. > tapply (my_df$height, my_df$gender,mean) 
  23.        F        M 
  24. 175.5000 181.3333 
  26. > class(tapply (my_df$height, my_df$gender,mean) )
  27. [1] "array"
  1. > library(reshape2)
  2. > head(tips)
  3.   total_bill  tip    sex smoker day   time size
  4. 1      16.99 1.01 Female     No Sun Dinner    2
  5. 2      10.34 1.66   Male     No Sun Dinner    3
  6. 3      21.01 3.50   Male     No Sun Dinner    3
  7. 4      23.68 3.31   Male     No Sun Dinner    2
  8. 5      24.59 3.61 Female     No Sun Dinner    4
  9. 6      25.29 4.71   Male     No Sun Dinner    4
  10. > ddply(tips,.(sex,smoker),function(x) sum(x$tip)/sum(x$total_bill))
  11.      sex smoker        V1
  12. 1 Female     No 0.1531892
  13. 2 Female    Yes 0.1630623
  14. 3   Male     No 0.1573122
  15. 4   Male    Yes 0.1369188
  16. > 
  17. > ddply(tips,~sex+smoker,function(x) sum(x$tip)/sum(x$total_bill)) #另一种写法
  18.      sex smoker        V1
  19. 1 Female     No 0.1531892
  20. 2 Female    Yes 0.1630623
  21. 3   Male     No 0.1573122
  22. 4   Male    Yes 0.1369188

dlply(iris,~Species,my_mode)适合回归分析,输入数据框，输出list

lapply() 返回list 回归分析返回的是list，所有lapply适合回归分析
dlply适合回归分析,输入数据框，输出list

  1. > my_mode = function(x) lm(Sepal.Length~Sepal.Width,data = x)
  3. > dlply(iris,~Species,my_mode) #iris是数据，~Species是分类变量，my_mode是函数输出的是list
  4. $setosa
  6. Call:
  7. lm(formula = Sepal.Length ~ Sepal.Width, data = x)
  9. Coefficients:
  10. (Intercept)  Sepal.Width  
  11.      2.6390       0.6905  
  14. $versicolor
  16. Call:
  17. lm(formula = Sepal.Length ~ Sepal.Width, data = x)
  19. Coefficients:
  20. (Intercept)  Sepal.Width  
  21.      3.5397       0.8651  
  24. $virginica
  26. Call:
  27. lm(formula = Sepal.Length ~ Sepal.Width, data = x)
  29. Coefficients:
  30. (Intercept)  Sepal.Width  
  31.      3.9068       0.9015  
  34. attr(,"split_type")
  35. [1] "data.frame"
  36. attr(,"split_labels")
  37.      Species
  38. 1     setosa
  39. 2 versicolor
  40. 3  virginica

彩蛋

each、colwise、numwise和ddply函数

  1. > each(mean,sd,median)(iris$Sepal.Length) #一次性的对某个对象（第二个括号）进行多种操作
  2.      mean        sd    median 
  3. 5.8433333 0.8280661 5.8000000 
  4. > colwise(mean)(iris)#对列进行操作
  5.   Sepal.Length Sepal.Width Petal.Length Petal.Width Species
  6. 1     5.843333    3.057333        3.758    1.199333      NA
  7. Warning message:
  8. In mean.default(X[[i]], ...) :
  9.   argument is not numeric or logical: returning NA
  10. > numcolwise(mean)(iris) #只对为数据的列进行操作
  11.   Sepal.Length Sepal.Width Petal.Length Petal.Width
  12. 1     5.843333    3.057333        3.758    1.199333
  13. > #ddply的好处
  14. > #写法一
  15. > ddply(iris,~Species,colwise(mean,c('Sepal.Length','Sepal.Width')))
  16.      Species Sepal.Length Sepal.Width
  17. 1     setosa        5.006       3.428
  18. 2 versicolor        5.936       2.770
  19. 3  virginica        6.588       2.974
  20. > #写法二
  21. > ddply(iris,~Species,colwise(mean,.(Sepal.Length,Sepal.Width)))
  22.      Species Sepal.Length Sepal.Width
  23. 1     setosa        5.006       3.428
  24. 2 versicolor        5.936       2.770
  25. 3  virginica        6.588       2.974
  26. > #写法三
  27. > ddply(iris,~Species,colwise(mean,~Sepal.Length+Sepal.Width))
  28.      Species Sepal.Length Sepal.Width
  29. 1     setosa        5.006       3.428
  30. 2 versicolor        5.936       2.770
  31. 3  virginica        6.588       2.974

参考总结图
image.png