R 聚合 data.frame 一列中的日期和时间格式错误

Question

我有一个如下所示的数据框：

   kWh   Equipment                date
1 1.53 aquecedor01 2015-01-01 00:00:00
2 5.29 aquecedor01 2015-01-01 01:00:00
3 5.73 aquecedor01 2015-01-01 02:00:00

但是当我通过 Equipment 变量汇总数据以从 kWh 中找到最大值时，日期列的格式错误如下：

     Equipment kWh       date
1  aquecedor01 6.5 1433023200
2  aquecedor02 6.5 1433023200
3     exaustor 6.5 1433023200

我已经为此苦苦挣扎了一段时间，我发现的大多数东西只能独立地处理日期或时间。就我而言，因为我是在 Shiny 应用程序中执行情节，所以一次完成所有操作对我来说会更容易。

我想在条形图中绘制每个设备的所有最大值，并在条形图上写下该值的时间。这是我的代码：

ggplotly(ggplot(data=aggregate(
        . ~ Equipment,
        data = dt.hourly[(as.character(input$dateRange[1]) <= dt.hourly$date) & 
                           (as.character(input$dateRange[2]) > dt.hourly$date) &
                           (dt.hourly$Equipment %in% input$equipments),], max),
        aes(x=Equipment, y=kWh)) +
        geom_bar(position = 'dodge', stat='identity') +
        geom_text(aes(label=date),
                  position = position_stack(vjust = 0.5),
                  angle = 90,
                  size=2) +
        xlab("Date") + 
        ylab("Consumption (kWh)") +
    theme(axis.text.x = element_text(angle = 90, hjust = 1))
    ))

除此之外 angle=90 被忽略了，我不明白为什么。这就是我得到的：

提前致谢。

作为一个可重现的例子：

library(plotly)
set.seed(1)
dt <- data.frame(
  kWh = sample(10:100, 10, replace = TRUE)/100,
  Equipment = sample(c("heater", "furnace", "AC"), 10, replace = TRUE),
  date = sample(as.POSIXct(c("2015-01-14 17:00:00", "2015-01-21 20:00:00", "2015-01-21 22:00:00", "2015-02-21 20:00:00", "2015-01-22 14:00:00", "2015-02-14 17:00:00", "2015-02-21 20:00:00", "2015-02-21 22:00:00", "2015-03-21 20:00:00", "2015-03-22 14:00:00" )), 10, replace = TRUE)
)

以及绘图：

ggplotly(ggplot(data=aggregate(
  . ~ Equipment,
  data = dt[("2015-01-12" <= dt$date) & 
              ("2015-02-22" > dt$date) &
              (dt$Equipment %in% c("AC", "furnace")),], max),
  aes(x=Equipment, y=kWh)) +
  geom_bar(position = 'dodge', stat='identity') +
  geom_text(aes(label=date),
            position = position_stack(vjust = 0.5),
            angle = 90,
            size=2) +
  xlab("Date") + 
  ylab("Consumption (kWh)") +
  theme(axis.text.x = element_text(angle = 90, hjust = 1)))

并且 dput 输出是：

structure(list(kWh = c(0.34, 0.43, 0.62, 0.92, 0.28, 0.91, 0.95, 
0.7, 0.67, 0.15), Equipment = structure(c(3L, 3L, 1L, 2L, 1L, 
2L, 1L, 1L, 2L, 1L), .Label = c("AC", "furnace", "heater"), class = "factor"), 
    date = structure(c(1427032800, 1421877600, 1424548800, 1421870400, 
    1421877600, 1424548800, 1421254800, 1424548800, 1426968000, 
    1424548800), class = c("POSIXct", "POSIXt"), tzone = "")), class = "data.frame", row.names = c(NA, 
-10L))

Answer 1

由于您的 objective 是注释出现最大 kWh 的日期，因此您希望在中省略 date聚合。因此，考虑使用 ave 计算分组 max_kWh，这会添加相同长度的列（内联聚合）。然后在 kWh == max_kWh.

处对数据框进行子集化

dt$max_kWh <- with(dt, ave(kWh, Equipment, FUN=max))

agg_dt <- subset(dt, kWh == max_kWh)

ggplot(data=agg_dt, aes(x=Equipment, y=kWh)) +
  geom_bar(position = 'dodge', stat='identity') +
  geom_text(aes(label=date),
            position = position_stack(vjust = 0.5),
            angle = 0,
            size = 2) +
  xlab("Equipment") + 
  ylab("Consumption (kWh)") +
  theme(axis.text.x = element_text(angle = 0, hjust = 1))

对于读取 input 值的 Shiny 集成，使用 transform 添加 max_kWh 列，然后将结果包装在 subset:

agg_dt <- subset(
             transform(dt.hourly[(as.character(input$dateRange[1]) <= dt.hourly$date) & 
                                 (as.character(input$dateRange[2]) > dt.hourly$date) &
                                 (dt.hourly$Equipment %in% input$equipments),],
                       max_kWh = ave(kWh, Equipment, FUN=max),
             kWh == max_kWh
          )

ggplotly(ggplot(data=agg_dt, aes(x=Equipment, y=kWh)) +
            geom_bar(position = 'dodge', stat='identity') +
            geom_text(aes(label=date),
                      position = position_stack(vjust = 0.5),
                      angle = 0,
                      size = 2) +
            xlab("Date") + 
            ylab("Consumption (kWh)") +
        theme(axis.text.x = element_text(angle = 0, hjust = 1))
))

Answer 2

您可以在绘制之前根据需要过滤数据：

library(tidyverse) 

dt_sum <- dt %>% 
  # First filter according to your input 
  filter(Equipment %in% c("AC", "furnace") & ("2015-01-12" <= date) & ("2015-02-22" > date)) %>%  
  group_by(Equipment) %>%   #  Group the data by Equipment
  top_n(1, kWh) %>%   # Take the maximum kWh value per Equipment
  top_n(1, date)      # Take the maximum date if there are several with the same max kWh value

dt_sum
# A tibble: 2 x 3
# Groups:   Equipment [2]
#     kWh Equipment date               
#   <dbl> <fct>     <dttm>             
# 1  0.92 furnace   2015-01-21 20:00:00
# 2  0.95 AC        2015-01-14 17:00:00

p <- ggplot(dt_sum, aes(x = Equipment, y = kWh)) +
  geom_bar(position = 'dodge', stat = 'identity') +
  geom_text(aes(label = date), position = position_stack(vjust = 0.5),
            angle = 90, size = 2) +
  xlab("Date") +
  ylab("Consumption (kWh)") +
  theme(axis.text.x = element_text(angle = 90, hjust = 1))

p

角度问题是由 ggplotly 引起的（如您所见，angle = 90 在 ggplot 调用中未被忽略）。

ggplotly(p)

R 聚合 data.frame 一列中的日期和时间格式错误

R aggregate data.frame having dates and hours in one column misformatted

aggregate

r

ggplot2

shiny

ggplotly