#变量的重编码
> #在分析数据时我们经常会遇到将变量值转换成其他的值的情况（如：将连续变量转成分类变量）这时
> #时我们就需要我们对原有数据进行重新编码。本文将介绍R软件中常用的三种编码方法
> #1，使用逻辑判断式编码
> #2，使用cut函数编码
> #3,使用car程序包的recode函数
> #(1)使用逻辑判断式编码
> #（1）现假设我们需要将下面的连续型变量x按照10与20分成三个组，新的分组名称为1、2、3：
> x <- c(4,12,50,18,50,22,23,46,8,46,36,18,10,14,35,48,23,17,29,30)
> x2 <- 1*(x <=10) + 2*(x>10 &x<=20) + 3*(x>20)
> #将上述变量的数字编码改为字符编码
> #利用变量x2中的数值作为另一个变量的位置信息这样不同的值就取到另一个字符变量不同的变量
> #这样就变成了字符编码
> labels <- c("A","B","C")
> x3 <- labels[x2]
> x2
 [1] 1 2 3 2 3 3 3 3 1 3 3 2 1 2 3 3 3 2 3 3
> x3
 [1] "A" "B" "C" "B" "C" "C" "C" "C" "A" "C" "C" "B" "A" "B" "C" "C" "C" "B" "C" "C"
> #列2
> income <- c(130065,82961,133076,123028,108945,173466,17477)
> income
[1] 130065  82961 133076 123028 108945 173466  17477
> newcodes <- c("低收入","中等收入","高收入")
> index=1*(income < 20000) + 2*(income >=20000 & income <= 60000) + 3*(income >60000)
> income = newcodes[index]
> income
[1] "高收入" "高收入" "高收入" "高收入" "高收入" "高收入" "低收入"
> #(2)使用ifelse函数
> #基本语法：ifelse(逻辑判断式，TRUE - 表达式,FALSE-表达式)
> #编码分为两个组
> x 
 [1]  4 12 50 18 50 22 23 46  8 46 36 18 10 14 35 48 23 17 29 30
> #(2)使用ifelse函数
> #基本语法：ifelse(逻辑判断式，TRUE - 表达式,FALSE-表达式)
> #编码分为两个组
> x <- c(4,12,50,18,50,22,23,46,8,46,36,18,10,14,35,48,23,17,29,30)
> #使用ifelse函数满足条件返回1，不满足条件返回2
> x2=ifelse(x<=30,1,2)
> x2
 [1] 1 1 2 1 2 1 1 2 1 2 2 1 1 1 2 2 1 1 1 1
> x3=ifelse(x<=30,"A","B")
> #搭配%int%运算符，将"A","C"重编码为"Group1","B","D"重编码为Group2
> y <- c("B","A","C","C","B","A","D","B","C","D")
> y2 <- ifelse(y %in% c("A","C"),"Group1","Group2")
> y2
 [1] "Group2" "Group1" "Group1" "Group1" "Group2" "Group1" "Group2" "Group2" "Group1"
[10] "Group2"
> #当编码成三个或者三个以上的组时需要多次使用ifelse函数（相当于Excel中的嵌套if函数）
> #将x按照10与20两个分隔点分成1，2，3三组
> x <- c(4,12,50,18,50,22,23,46,8,46,36,18,10,14,35,48,23,17,29,30)
> x2 <- ifelse(x<=10,1,ifelse(x>20,3,2))
> x2
 [1] 1 2 3 2 3 3 3 3 1 3 3 2 1 2 3 3 3 2 3 3
> y2 <- ifelse(y %in% c("A","E"),1,ifelse(y %in% "C",2,3))
> y2
 [1] 3 1 2 2 3 1 3 3 2 3
 >二，使用 cut()函数编码
> #其中
> #x为数值向量
> #breaks为分割点信息。若breaks为向量，则根据向量中的数字进行分割，若breaks为大于1的正整数k，
> #则将x分为均等的k组
> #labels为分割后各组的名称，若为null，则输出为数字向量，否则输出factor变量。
> #include.lowest=FALSE表示分割时不含各区间端点的最小值
> #right=T表示各区间左端为open，右端为closed的区间
> ?cut
> x2 <- cut(x,breaks = C(0,10,20,max(x)))
Error in C(0, 10, 20, max(x)) : object not interpretable as a factor
> x2 <- cut(x,breaks = C(0,10,20,max(x)),labels = c(1,2,3))
Error in C(0, 10, 20, max(x)) : object not interpretable as a factor
> x2 <- cut(x,breaks = C(0,10,20,48),labels = c(1,2,3))
Error in C(0, 10, 20, 48) : object not interpretable as a factor
> x2 <- cut(x,breaks = c(0,10,20,max(x)),labels = c(1,2,3))
> x2
 [1] 1 2 3 2 3 3 3 3 1 3 3 2 1 2 3 3 3 2 3 3
Levels: 1 2 3
> x2 <- cut(x,breaks = c(0,10,20,max(x)))
> x2
 [1] (0,10]  (10,20] (20,50] (10,20] (20,50] (20,50] (20,50] (20,50] (0,10]  (20,50]
[11] (20,50] (10,20] (0,10]  (10,20] (20,50] (20,50] (20,50] (10,20] (20,50] (20,50]
Levels: (0,10] (10,20] (20,50]
> x2 <- cut(x,breaks = c(0,10,20,max(x)),labels = c(1,2,3),include.lowest = TRUE)
> x2
 [1] 1 2 3 2 3 3 3 3 1 3 3 2 1 2 3 3 3 2 3 3
Levels: 1 2 3
> x2 <- cut(x,breaks = c(0,10,20,max(x)),include.lowest = TRUE)
> x2
 [1] [0,10]  (10,20] (20,50] (10,20] (20,50] (20,50] (20,50] (20,50] [0,10]  (20,50]
[11] (20,50] (10,20] [0,10]  (10,20] (20,50] (20,50] (20,50] (10,20] (20,50] (20,50]
Levels: [0,10] (10,20] (20,50]
> x2 <- cut(x,breaks = c(0,10,20,max(x)),labels = c(1,2,3),include.lowest = TRUE)
> x2
 [1] 1 2 3 2 3 3 3 3 1 3 3 2 1 2 3 3 3 2 3 3
Levels: 1 2 3
> as.vector(x2)
 [1] "1" "2" "3" "2" "3" "3" "3" "3" "1" "3" "3" "2" "1" "2" "3" "3" "3" "2" "3" "3"
> score
Error: object 'score' not found
> #现在我们模拟产生10个N(60,10)的随机成绩，并使用cut函数的breaks选项将其分成5组
> #生成10个平均值为60，标准差为10的正太分布数值并取整
> score = round(rnorm(10,60,10))
> score
 [1] 41 58 67 75 73 64 81 75 66 61
> score <- cut(score,breaks = 5)
> score
 [1] (41,49] (57,65] (65,73] (73,81] (65,73] (57,65] (73,81] (73,81] (65,73] (57,65]
Levels: (41,49] (49,57] (57,65] (65,73] (73,81]
> #由以上结果可知，cut()函数默认输出一个factor变量，并且自动将五个分组命名为（39，46.2] ...(67.8,75]
> cut()函数返回的分组标签名称有三种方式，第一种通过参数labels主动设置标签名称，第二种使用cut（）函数的默认值返回区间作为标签的名称如：(41,49] (57,65] (65,73] (73,81] (65,73] 这种类型的，第三种设置labels=FALSE则返回的标签名称是数值所在的第几区间如：5 3 4 3 5 3 5 1 3 4，其中1表示最大的区间，5表示最小的区间。
> #如何cut()的选项labels=FALSE，则输出的结果是数字编码（返回在第几个区间）
> score = round(rnorm(10,60,10))
> score.cut <- cut(score,breaks = 5,labels = FALSE)
> score.cut
 [1] 5 3 4 3 5 3 5 1 3 4
> score.cut = cut(score,breaks = 5 )
> score.cut
 [1] (60.2,64]   (52.6,56.4] (56.4,60.2] (52.6,56.4] (60.2,64]   (52.6,56.4] (60.2,64]  
 [8] (45,48.8]   (52.6,56.4] (56.4,60.2]
Levels: (45,48.8] (48.8,52.6] (52.6,56.4] (56.4,60.2] (60.2,64],
>三,使用car程序包的recode函数
> #recodes参数的值是一个字符串，字符串里面是以分号分隔的编码规则：
> #recodes=“规则1","规则2",..."
> #每一个编码规则的格式为旧码列表=新码，“旧码列表”部分可用lo代表旧码的最小值，
> #hi代表旧码的最大值，撰写规则如下
> #（1）旧码=新码 旧码只有一个单数值。例如："0=NA"表示将0改为NA。
> #（2）旧码向量=新码 多个旧码改为一个新码。例如："c(7,8,9)="hight"，将7，8，9改为high
> #（3）start:end=新码 有序数字改码。例如"lo:19="c".
> #（4）else=新码 所有其他情况。例如："else=NA".
> recodes与ifelse函数比较类似，但是在需要分成多组的情况下，recodes更好用（不用ifelse函数多层嵌套），相对于cut函数更加灵活，可以应对更加复杂的分组方式，如把等于A或者C的分到一个组这种情况。
> #例子
> library(carData)
> library(car)
> x2
 [1] 1 2 3 2 3 3 3 3 1 3 3 2 1 2 3 3 3 2 3 3
Levels: 1 2 3
> x <- c(1,2,3,1,2,3,1,2,3)
> recode(x,"c(1,2)='A';else='B'")
[1] "A" "A" "B" "A" "A" "B" "A" "A" "B"
> #将成绩0~40分之间的分数编码为1，41-60分之间为2，61-80分为3，81以上为4，其他情况为NA
> score
 [1] 62 55 58 56 64 56 61 45 55 57
> recode(score,"0:40=1;41:60=2;61:80=3;81:hi=4;else=NA")
 [1] 3 2 2 2 3 2 3 2 2 2
> #上例子改为‘A' 'B' 'C' 'D'
> recode(score,"0:40='A';41:60='B';61:80='C';81:hi='D';else=NA")
 [1] "C" "B" "B" "B" "C" "B" "C" "B" "B" "B"