R rbind行到df，结果按名称按组计算多列

Question

如何按 "id" 分组，将一些算术函数应用于最后四列（按组），并将新行添加到包含结果的 df。这是一个包含 5 个样本 (id) 和 8 列的说明性示例：

    d1   d2   id  type         treat  v1_gm  v2_pct v3_pct
1   info info 1   leaf         NA     0.2    70     90
2   info info 1   flower       A      0.5    80     80
3   info info 2   leaf         NA     0.4    65     80
4   info info 2   flower       A      0.1    90     90
5   info info 3   leaf         NA     0.6    55     80
6   info info 3   stem         A      0.3    80     30
7   info info 4   leaf         NA     0.6    30     40
8   info info 4   flower       A      0.7    75     75
9   info info 5   leaf/stem    NA     0.8    80     75

可重现的例子：

df <- data.frame(matrix(NA, nrow = 9, ncol = 8), row.names=NULL)
colnames(df) <- c("d1","d2","id","type","treat","v1_gm","v2_pct","v3_pct")
df$d1 <- "info"
df$d2 <- "info"
id <- c(1,1,2,2,3,3,4,4,5)
df$id <- c(1,1,2,2,3,3,4,4,5)
df$type <- c("leaf","flower","leaf","flower","leaf","stem","leaf","flower","leaf/stem")
df$treat <- c(NA,"A",NA,"A",NA,"A",NA,"A",NA)
df$v1_gm <- c(0.2,0.5,0.4,0.1,0.6,0.3,0.6,0.7,0.8)
df$v2_pct <- c(70,80,65,90,55,80,30,75,80)
df$v3_pct <- c(90,80,80,90,80,30,40,75,75)

结果 table 应如下所示。第 3、6、9 和 13 行是包含结果的新行。新行可以附加在 table 的末尾，或者放入一个 tmp df 以便稍后使用 rbind 添加（无论哪种方式，我都不知道该怎么做）。分组变量是列 "id"。函数 sum 用于 "v1_gm"。函数 "mean" 用于多个连续的列，此处为 "v1_pct" 和 "v3_pct"，应按名称调用（例如，v1_pct:v3_pct）。新行中 "type" 的值是从组行中的 "type" 连接而来的， "d1" 和 "d2" 只是从 treat==[=31 的组行中复制而来=]，并且新行中的 "treat" 被赋值为 "cmb"。

    d1   d2   id type         treat v1_gm  v2_pct  v3_pct
1   info info 1  leaf         NA    0.2    70      90
2   info info 1  flower       A     0.5    80      80
3   info info 1  leaf/flower  cmb   0.7    75      85
4   info info 2  leaf         NA    0.4    65      80
5   info info 2  flower       A     0.1    90      90
6   info info 2  leaf/flower  cmb   0.5    77.5    85
7   info info 3  leaf         NA    0.6    55      80
8   info info 3  stem         A     0.3    80      30
9   info info 3  leaf/stem    cmb   0.9    67.5    55
10  info info 4  leaf         NA    0.6    30      40
11  info info 4  flower       A     0.7    75      75
13  info info 4  leaf/flower  cmb   1.3    52.5    57.5
14  info info 5  leaf/stem    NA    0.8    80      75

Answer 1

我不确定您是否可以将组摘要作为一行添加到数据框中。你应该可以把它作为一个专栏来做。

library("dplyr")
res1 <- df %>% group_by(id) %>% mutate( sumV1 = sum(v1_gm),meanV2 = mean(v2_pct),meanV3 = mean(v3_pct),gr_type = paste(type,collapse="/")) %>% filter(treat == "A") %>% select(d1,d2,id,type,v1_gm=sumV1, v2_pct = meanV2, v3_pct = meanV3,type = gr_type)

这会给你答案然后使用bind_rows你会得到你想要的结果

final_res <- bind_rows(df,res1)

Answer 2

通过对@Arun 的回答进行一些修改，以下脚本完全解决了这个问题。

library("dplyr")
res1 <- df %>%  
  group_by(id) %>%  
  mutate(  
    v1_gm = sum(v1_gm),  
    v2_pct = mean(v2_pct),  
    v3_pct = mean(v3_pct),  
    type = paste(type,collapse="/")) %>%  
  filter(treat == "A") %>%  
  mutate(treat = as.character("calculated"))  
final_res1 <- bind_rows(df,res1)  
final_res1$id <- as.character(final_res1$id)  
final_res1 <- final_res1 [order(final_res1$id, final_res1$treat, na.last=FALSE),  ]

R rbind行到df，结果按名称按组计算多列

R rbind rows to df with results calculated by group for multiple columns by name

r

group-summaries

dplyr