R クロス集計（acast）

目的：Rでエクセルのピポッドテーブルのようなクロス集計をおこなう
方法：acastを使う
補足：library(reshape2)が必要

 #サンプルデータ作成
v.x1 <- c("oda","oda","oda","toyo","toyo","toyo","ie","ie","ie","ie")
v.x2 <- c("a","a","b","b","b","c","c","c","c","c")
v.x3 <- c(1000,900,800,700,600,500,400,300,200,100)
df.x <- data.frame(user_id = v.x1, item = v.x2, price =v.x3)

df.x

> df.x
   user_id item price
1      oda    a  1000
2      oda    a   900
3      oda    b   800
4     toyo    b   700
5     toyo    b   600
6     toyo    c   500
7       ie    c   400
8       ie    c   300
9       ie    c   200
10      ie    c   100

 #クロス集計
tmp <- acast(df.x, user_id ~ item, sum, value.var = "price")

tmp

> tmp
        a    b    c
ie      0    0 1000
oda  1900  800    0
toyo    0 1300  500

#補足：user_idの列を無理やり作る

df.cross <- data.frame(tmp)
write.csv(df.cross,file = "./cross.csv")
df.cross <- fread("./cross.csv") #llibrary(data.table)が必要
df.cross <- data.frame(df.cross)

df.cross

> df.cross
    V1    a    b    c
1   ie    0    0 1000
2  oda 1900  800    0
3 toyo    0 1300  500

names(df.cross)[1] <- c("user_id")

df.cross

> df.cross
  user_id    a    b    c
1      ie    0    0 1000
2     oda 1900  800    0
3    toyo    0 1300  500

追記：acastは遅いのでspreadがお勧め
R　高速に大規模データのクロス集計をおこなう（tally, spread）