R - 根据其他列查找三个最高值

Question

我有一个数据框（但也可以是数据表），其中包含多个站点（其余列“a”到“l”）中植被类型的比例（第一列“type”）。这是它的样子：

dat <- structure(list(type = structure(1:17, class = "factor", .Label = c("not_vegetated", 
"needleleaf_evergreen_temperate_tree", "needleleaf_evergreen_boreal_tree", 
"needleleaf_deciduous_boreal_tree", "broadleaf_evergreen_tropical_tree", 
"broadleaf_evergreen_temperate_tree", "broadleaf_deciduous_tropical_tree", 
"broadleaf_deciduous_temperate_tree", "broadleaf_deciduous_boreal_tree", 
"broadleaf_evergreen_shrub", "broadleaf_deciduous_temperate_shrub", 
"broadleaf_deciduous_boreal_shrub", "c3_arctic_grass", "c3_non.arctic_grass", 
"c4_grass", "c3_crop", "c3_irrigated")), a = c("0", "1.55", "0", 
"0", "0", "0", "0", "11.59", "0", "0", "0", "0", "0", "31.5", 
"0.26", "52.29", "0"), b = c("0", "8.27", "0.02", "0", "0", "0", 
"0", "35.5", "0.05", "0", "0.04", "0", "0", "26.02", "0", "15.7", 
"0"), c = c("1.42", "7.55", "0", "0", "0", "0", "0", "14.24", 
"2.38", "0", "0.06", "0", "0", "31.79", "0", "36.56", "0"), d = c("0", 
"13.87", "3.97", "0", "0", "0", "0", "51.66", "7.68", "0", "0.07", 
"0", "0", "8.18", "0", "10.23", "0"), e = c("0", "16.23", "0.24", 
"0", "0", "0", "0", "67.15", "2.12", "0", "0", "0", "0", "6.52", 
"0.1", "4.68", "0"), f = c("0.26", "6.29", "60.98", "0", "0", 
"0", "0", "3.72", "10.49", "0", "0", "0.3", "2.45", "3.61", "0", 
"3.9", "0"), g = c("0.11", "38.16", "10.9", "0", "0", "0", "0", 
"31.72", "13.53", "0", "0", "0", "0", "0.68", "0", "0.9", "0"
), h = c("0", "10.42", "0.42", "0", "0", "0", "0", "55.44", "1.49", 
"0", "0", "0", "0", "16.54", "0", "13.33", "0"), i = c("0", "1.39", 
"0", "0", "0", "0", "0", "11.56", "0", "0", "0", "0", "0", "33.94", 
"0", "49.26", "0"), j = c("0.45", "16.48", "0", "0", "0", "0", 
"0", "40.02", "0", "0", "0.02", "0", "0", "17.53", "0", "10.51", 
"0"), k = c("6.02", "2.78", "0", "0", "0", "0", "0", "1.27", 
"2.51", "0", "0.03", "0", "0", "18.29", "0", "69.1", "0"), l = c("6.13", 
"22.77", "2.72", "0", "0", "0", "0", "22.69", "3.85", "0", "0.06", 
"0", "0", "1.25", "0", "1.53", "0")), .Names = c("type", 
"a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l"), row.names = c(NA, 
-17L), class = "data.frame")

我需要做的是计算每个站点的三个更丰富的植被类型分数，并将其汇总在与此类似的数据框（或数据表）中（为简洁起见仅显示前两个站点）：

            a                                                   b
c3_crops (52.29)                                broadleaf_deciduous_temperate_tree (35.5)
c3_non.arctic_grass (31.5)                      c3_non.arctic_grass (26.02)
broadleaf_deciduous_temperate_tree (11.59)      c3_crop (15.7)

关于如何做到这一点的任何提示？

Answer 1

这是适合您的一种方法。如果我没看错你的问题，你想选择三个数字最高的三个站点。您首先要以长格式格式化数据，将字符转换为数字，然后按site 和value 对数据进行排序。然后，您通过site 定义组，并为每个组获取前三行。

library(dplyr)
library(tidyr)

gather(dat, key = site, value = value, -type) %>%
mutate(value = as.numeric(value)) %>%
arrange(site, desc(value)) %>%
group_by(site) %>%
top_n(3)

   type                               site  value
   <fct>                              <chr> <dbl>
 1 c3_crop                            a      52.3
 2 c3_non.arctic_grass                a      31.5
 3 broadleaf_deciduous_temperate_tree a      11.6
 4 broadleaf_deciduous_temperate_tree b      35.5
 5 c3_non.arctic_grass                b      26.0
 6 c3_crop                            b      15.7
 7 c3_crop                            c      36.6
 8 c3_non.arctic_grass                c      31.8
 9 broadleaf_deciduous_temperate_tree c      14.2
10 broadleaf_deciduous_temperate_tree d      51.7
# ... with 26 more rows

Answer 2

考虑将您的数据从宽调整为长，并将by 与order 和head(...,3) 全部使用基数R。by 下方返回数据帧的命名列表。

# RESHAPE DATA
rdat <- reshape(dat, timevar = "site", times = letters[1:12], varying = list(2:13), 
                v.names = "value", new.row.names = 1:1000, direction = "long")

rdat$value <- as.numeric(rdat$value)

# ORDER DATA
rdat <- with(rdat, rdat[order(site, -value),])  

# SUBSET BY SITE, RETURN TOP 3 ROWS
by(rdat, rdat$site, FUN=function(i) head(i, 3))

# rdat$site: a
#                                  type site value id
# 16                            c3_crop    a 52.29 16
# 14                c3_non.arctic_grass    a 31.50 14
# 8  broadleaf_deciduous_temperate_tree    a 11.59  8
# --------------------------------------------------------------------------------
# rdat$site: b
#                                  type site value id
# 25 broadleaf_deciduous_temperate_tree    b 35.50  8
# 31                c3_non.arctic_grass    b 26.02 14
# 33                            c3_crop    b 15.70 16
# --------------------------------------------------------------------------------
# rdat$site: c
#                                  type site value id
# 50                            c3_crop    c 36.56 16
# 48                c3_non.arctic_grass    c 31.79 14
# 42 broadleaf_deciduous_temperate_tree    c 14.24  8
# ...