传播/收集错误：必须提供符号或字符串作为参数

Question

我已经传播数据(df)

       Timestamp    area    count   type
2019-08-28 00:30:00 area1     4     A
2019-08-28 00:30:01 area1     1     B
2019-08-28 00:30:02 area1     8     C
2019-08-28 00:30:03 area2     8     A
2019-08-28 00:30:04 area2     1     B
2019-08-28 00:30:04 area2     8     C
2019-08-28 00:30:06 area3     18    A

进入

      Timestamp       area    A    B    C


  2019-08-28 00:30:00 area1   4   NA  NA
  2019-08-28 00:30:01 area1   NA  NA  1
  2019-08-28 00:30:02 area1   NA  8   NA
  2019-08-28 00:30:03 area2   8   NA  NA
  2019-08-28 00:30:04 area2   NA  NA  1
  2019-08-28 00:30:04 area2   NA  8   NA
  2019-08-28 00:30:06 area3   18  NA  NA

使用代码

 df %>%
group_by(Timestamp, area) %>%
  spread(type, count)

然后，我想将同一区域的相邻行合并，使表格看起来像这样

        Timestamp     area    A    B    C
      <S3: POSIXct>   <chr>  <dbl><dbl><dbl>

  2019-08-28 00:30:00 area1   4   8   1


  2019-08-28 00:30:03 area2   8   8   1


  2019-08-28 00:30:06 area3   18  ... ...
.
.

我用过代码

df %>%
  gather(type, count,-area) %>%
  group_by(type) %>%
  mutate(Rank = dense_rank(type), 
         RankOrder = paste0("ord_", row_number())) %>%
  spread(type) %>%
  spread(RankOrder, Rank) %>%
  summarise_all(funs(.[which(!is.na(.))]))

但我得到了错误

eval_tidy(enquo(var), var_env) 中的错误：对象“未找到”。

要强调的一点：

每天都会重复记录数据，即。 area 每天都会重复。因此我们需要按天分组（不是日期时间，我只想提取日期）然后按区域分组我猜？但我不知道该怎么做。

最终目标是绘制type中各个元素之间的相关图。

请告诉我如何做到这一点。非常感谢。

Answer 1

以下方法的工作原理是将数据保持为长格式，直到您想在最后以宽格式查看它。基本做法是：

library(dplyr)
library(tidyr)
library(lubridate)

df <- tribble(
~Timestamp, ~area, ~count, ~type,
"2019-08-28 00:30:00", "area1", 4, "A",
"2019-08-28 00:30:01", "area1", 1, "B",
"2019-08-28 00:30:02", "area1", 8, "C",
"2019-08-28 00:30:03", "area2", 8, "A",
"2019-08-28 00:30:04", "area2", 1, "B",
"2019-08-28 00:30:04", "area2", 8, "C",
"2019-08-28 00:30:06", "area3", 18, "A")

df$Timestamp <- ymd_hms(df$Timestamp)
df$date <- ymd_hms(df$Timestamp) %>% date()
df$area <- factor(df$area)
df$type <- factor(df$type)

df %>%
  group_by(date, area, type) %>%
  summarize(count = sum(count)) %>%
  spread(key = type, value = count)

# # A tibble: 3 x 5
# # Groups:   date, area [3]
# date       area      A     B     C
# <date>     <fct> <dbl> <dbl> <dbl>
# 2019-08-28 area1     4     1     8
# 2019-08-28 area2     8     1     8
# 2019-08-28 area3    18    NA    NA

Answer 2

我们可以首先将数据spread 转换为宽格式，对于每个area，如果列中的所有值都是NA，我们返回NA，或者返回第一个非NA 值。

library(dplyr)

df %>%
  tidyr::spread(type, count) %>%
  group_by(area) %>%
  summarise_all(~if (all(is.na(.))) NA else first(.[!is.na(.)]))

#  area  Timestamp               A     B     C
#  <fct> <dttm>              <int> <int> <int>
#1 area1 2019-08-28 00:30:00     4     1     8
#2 area2 2019-08-28 00:30:03     8     1     8
#3 area3 2019-08-28 00:30:06    18    NA    NA

数据

df <- structure(list(Timestamp = structure(c(1566923400, 1566923401, 
1566923402, 1566923403, 1566923404, 1566923404, 1566923406), class = c("POSIXct", 
"POSIXt"), tzone = ""), area = structure(c(1L, 1L, 1L, 2L, 2L, 
2L, 3L), .Label = c("area1", "area2", "area3"), class = "factor"), 
count = c(4L, 1L, 8L, 8L, 1L, 8L, 18L), type = structure(c(1L, 
2L, 3L, 1L, 2L, 3L, 1L), .Label = c("A", "B", "C"), class = "factor")), 
row.names = c(NA, -7L), class = "data.frame")