R - 来自 Reshape2 的 dcast - 按值变量而不是变量名排序答案

【问题标题】：R - dcast from Reshape2 - order by value variable and not variable nameR - 来自 Reshape2 的 dcast - 按值变量而不是变量名排序
【发布时间】：2017-08-23 08:16:45
【问题描述】：

我有以下名为countries_tools 的data.frame。它由 3 列组成（一个日期时间列（过去 13 个月）、一个名称列（包含国家/地区）和一个访问列（来自这些特定国家/地区的访问该页面的人））：

datetime            name           Visits
2016-07-01 00:00:00 China          5237
2016-07-01 00:00:00 Germany        1434
2016-07-01 00:00:00 United States  1530
2016-07-01 00:00:00 India          696
2016-07-01 00:00:00 Japan          569
...
2017-07-01 00:00:00 China          4484
2017-07-01 00:00:00 Germany        1593
2017-07-01 00:00:00 United States  1438
2017-07-01 00:00:00 India          1204
2017-07-01 00:00:00 Japan          538

请注意，我删除了中间的其他 11 个月。另请注意，该名称始终是相同 5 个国家/地区的列表，这与上一个分析月份（本例中为 2017 年 7 月）访问次数较多的 5 个国家相对应。

在此消息的末尾有一个 dput 与我的数据。

为了更好地了解跨月访问的数据和发展，我从我的 data.frame 中创建了一个dcast：

countries_tools <- dcast(countries_tools, datetime ~ name, value.var="Visits")

但是，生成的数据框会按国家/地区名称（按字母顺序）对列进行排序：

> names(countries_tools)
[1] "datetime"      "China"         "Germany"       "India"         "Japan"         "United States"

但是，我希望顺序是由值变量（访问）完成的，因此最佳顺序应该是：

日期时间、中国、德国、美国、印度、日本

可以做到吗（最好是不需要额外的步骤）？也可以使用其他功能。

数据

dput(countries_tools)
structure(list(datetime = structure(c(1467320400, 1467320400, 
1467320400, 1467320400, 1467320400, 1469998800, 1469998800, 1469998800, 
1469998800, 1469998800, 1472677200, 1472677200, 1472677200, 1472677200, 
1472677200, 1475269200, 1475269200, 1475269200, 1475269200, 1475269200, 
1477951200, 1477951200, 1477951200, 1477951200, 1477951200, 1480543200, 
1480543200, 1480543200, 1480543200, 1480543200, 1483221600, 1483221600, 
1483221600, 1483221600, 1483221600, 1485900000, 1485900000, 1485900000, 
1485900000, 1485900000, 1488319200, 1488319200, 1488319200, 1488319200, 
1488319200, 1490994000, 1490994000, 1490994000, 1490994000, 1490994000, 
1493586000, 1493586000, 1493586000, 1493586000, 1493586000, 1496264400, 
1496264400, 1496264400, 1496264400, 1496264400, 1498856400, 1498856400, 
1498856400, 1498856400, 1498856400), class = c("POSIXct", "POSIXt"
), tzone = "Europe/Moscow"), name = c("China", "Germany", "United States", 
"India", "Japan", "China", "Germany", "United States", "India", 
"Japan", "China", "Germany", "United States", "India", "Japan", 
"China", "Germany", "United States", "India", "Japan", "China", 
"Germany", "United States", "India", "Japan", "China", "Germany", 
"United States", "India", "Japan", "China", "Germany", "United States", 
"India", "Japan", "China", "Germany", "United States", "India", 
"Japan", "China", "Germany", "United States", "India", "Japan", 
"China", "Germany", "United States", "India", "Japan", "China", 
"Germany", "United States", "India", "Japan", "China", "Germany", 
"United States", "India", "Japan", "China", "Germany", "United States", 
"India", "Japan"), Visits = c(5237, 1434, 1530, 696, 569, 4422, 
1508, 1971, 672, 461, 3993, 1521, 1901, 2027, 517, 3656, 1764, 
1716, 993, 509, 5483, 3117, 2762, 1298, 594, 5548, 2804, 2365, 
1222, 551, 3747, 3083, 1917, 999, 496, 3903, 2136, 1751, 1229, 
611, 5638, 2721, 2074, 1569, 533, 4326, 1618, 1511, 1254, 458, 
4364, 2021, 1690, 1162, 462, 4462, 1572, 1517, 1068, 574, 4484, 
1593, 1438, 1204, 538)), .Names = c("datetime", "name", "Visits"
), row.names = c(NA, -65L), class = "data.frame")

【问题讨论】：

标签： r reshape2 dcast

【解决方案1】：

您可以将“名称”转换为有序因子，说明您想要的级别顺序：

countries_tools$name <- ordered(countries_tools$name, levels = unique(countries_tools$name))

现在可以了：

dcast(countries_tools, datetime ~ name, value.var="Visits")

【讨论】：