Question

我有基于行的迁移数据。

param <- c("A", "B", "C") 
df <- data.frame(Case1 = c("A", "A", "B", "B"), 
             Case2 = c("A", "C", "A", "B"), 
             Val = c(0.5,0.4,0.3,0.7))

所以这个数据框看起来像 Case1 Case2 Val 1 A A 0.5 2 A C 0.4 3 B A 0.3 4 B B 0.7 这种基于行的数据框应该在一种“迁移矩阵”中进行转换。

dd <- data.frame(cA = c(0.5, 0.3, 0), 
             cB = c(0, 0.7, 0), 
             cC = c(0.4,0,0)) 
rownames(dd) <- paste0("Case1","_", param) 
colnames(dd) <- paste0("Case2","_", param)

所以迁移矩阵看起来像

Case2_A Case2_B Case2_C Case1_A 0.5 0.0 0.4 Case1_B 0.3 0.7 0.0 Case1_C 0.0 0.0 0.0

有人知道在R中这样做的好方法吗？非常感谢你！

Answer 1

您可以使用dplyr和tidyr：

library(dplyr); library(tidyr)

df %>% 
       complete(Case1 = LETTERS[1:3], Case2 = LETTERS[1:3]) %>% 
       mutate_at(vars(starts_with("Case")), funs(paste("Case", ., sep = "_"))) %>% 
       spread(Case2, Val, fill = 0.0)

# Source: local data frame [3 x 4]

#   Case1 Case_A Case_B Case_C
#   <chr>  <dbl>  <dbl>  <dbl>
#1 Case_A    0.5    0.0    0.4
#2 Case_B    0.3    0.7    0.0
#3 Case_C    0.0    0.0    0.0

或者，如果您想具体保留列号：

df %>% 
      complete(Case1 = LETTERS[1:3], Case2 = LETTERS[1:3]) %>% 
      mutate(Case1 = paste('Case1', Case1, sep = "_"), 
             Case2 = paste('Case2', Case2, sep = "_")) %>% 
      spread(Case2, Val, fill = 0.0)

# Source: local data frame [3 x 4]

#     Case1 Case2_A Case2_B Case2_C
#     <chr>   <dbl>   <dbl>   <dbl>
# 1 Case1_A     0.5     0.0     0.4
# 2 Case1_B     0.3     0.7     0.0
# 3 Case1_C     0.0     0.0     0.0

Answer 2

以基地R：

df

  Case1 Case2 Val
1     A     A 0.5
2     A     C 0.4
3     B     A 0.3
4     B     B 0.7

library(reshape2)
levels(df$Case1) <- c(levels(df$Case1), 'C')
df <- dcast(df, Case1~Case2, value.var='Val', drop=FALSE)
rownames(df) <- paste('Case1', df[,1], sep='_')
df <- df[-1]
names(df) <- paste('Case2', names(df), sep='_')
df[is.na(df)] <- 0.0
df

     Case2_A Case2_B Case2_C
Case1_A     0.5     0.0     0.4
Case1_B     0.3     0.7     0.0
Case1_C     0.0     0.0     0.0

Answer 3

base R选项将前两个colunms转换为xtabs并factor从{{1} levels级别转换为unique后使用unlist选项修改了列，以便不删除某些组合。

Un1 <- sort(unique(unlist(df[1:2])))
df[1:2] <- lapply(df[1:2], factor, levels = Un1)
res <- xtabs(Val~Case1+Case2, df)

如果我们需要dimnames

dimnames(res) <- Map(paste, names(dimnames(res)), dimnames(res), MoreArgs = list(sep="_"))
names(dimnames(res)) <- NULL
res
#         Case2_A Case2_B Case2_C
#Case1_A     0.5     0.0     0.4
#Case1_B     0.3     0.7     0.0
#Case1_C     0.0     0.0     0.0

将基于行的迁移数据转换为迁移矩阵

3 个答案: