按名称汇总跨多个变量的数据
Posted
技术标签:
【中文标题】按名称汇总跨多个变量的数据【英文标题】：Summarizing data by name separated across multiple variables 【发布时间】：2018-12-26 13:51:04 【问题描述】：我正在尝试计算每位球员的进球总数、主要助攻数和次要助攻数。我的问题是我无法理解这样做的逻辑，因为我想通过（球员姓名）总结的数据列在三个变量（目标、主要助攻和次要助攻）中
这是我的可重现数据（来自dput()，因此为混乱道歉）。
mydata <- structure(list(primary_assist = c("Dmitry Gilyazitdinov", "Evgeny Orlov",
"Anton Burdasov", "Sergei Kalinin", "Stanislav Solovyov", "Vasily Streltsov",
NA, "Bogdan Potekhin", "Bogdan Potekhin", "Vasily Streltsov",
"Vasily Streltsov", "Viktor Postnikov", "Danil Kaskov", NA, NA,
"Artemy Panarin"), secondary_assist = c("Andrei Badrutdinov",
NA, NA, NA, "Danil Gubarev", "Nikita Manukhov", NA, "Evgeny Grigorenko",
"Daniil Apalkov", "Ivan Boiko", NA, "Viktor Antipin", "Vitaly Sychov",
NA, NA, "Stanislav Levin"), goal = c("Vitaly Kropachyov", "Dmitry Kozlov",
"Stanislav Solovyov", "Kirill Polyansky", "Anton Burdasov", "Ilya Solodov",
"Alexander Antropov", "Daniil Apalkov", "Evgeny Grigorenko",
"Alexander Antropov", "Alexander Antropov", "Evgeny Grigorenko",
"Denis Belonogov", "Vitaly Sychov", "Alexander Streltsov", "Pyotr Kopyttsov"
), team = c("Belye Medvedi", "Omskie Yastreby", "Belye Medvedi",
"Omskie Yastreby", "Belye Medvedi", "Avto", "Avto", "Stalnye Lisy",
"Stalnye Lisy", "Avto", "Avto", "Stalnye Lisy", "Avto", "Avto",
"Avto", "Russkie Vityazi"), game_strength = c("PP", "EV", "EV",
"EV", "EV", "PP", "SO", "EV", "PP", "PP", "EV", "PP", "PP", "EV",
"PP", "EV"), season = c("2009-10", "2009-10", "2009-10", "2009-10",
"2009-10", "2009-10", "2009-10", "2009-10", "2009-10", "2009-10",
"2009-10", "2009-10", "2009-10", "2009-10", "2009-10", "2009-10"
)), class = c("tbl_df", "tbl", "data.frame"), row.names = c(NA,
-16L), .Names = c("primary_assist", "secondary_assist", "goal",
"team", "game_strength", "season"))
mydata
#> # A tibble: 16 x 6
#> primary_assist secondary_assist goal team game_strength season
#> <chr> <chr> <chr> <chr> <chr> <chr>
#> 1 Dmitry Gilyazitdinov Andrei Badrutdin~ Vita~ Bely~ PP 2009-~
#> 2 Evgeny Orlov <NA> Dmit~ Omsk~ EV 2009-~
#> 3 Anton Burdasov <NA> Stan~ Bely~ EV 2009-~
#> 4 Sergei Kalinin <NA> Kiri~ Omsk~ EV 2009-~
#> 5 Stanislav Solovyov Danil Gubarev Anto~ Bely~ EV 2009-~
#> 6 Vasily Streltsov Nikita Manukhov Ilya~ Avto PP 2009-~
#> 7 <NA> <NA> Alex~ Avto SO 2009-~
#> 8 Bogdan Potekhin Evgeny Grigorenko Dani~ Stal~ EV 2009-~
#> 9 Bogdan Potekhin Daniil Apalkov Evge~ Stal~ PP 2009-~
#> 10 Vasily Streltsov Ivan Boiko Alex~ Avto PP 2009-~
#> 11 Vasily Streltsov <NA> Alex~ Avto EV 2009-~
#> 12 Viktor Postnikov Viktor Antipin Evge~ Stal~ PP 2009-~
#> 13 Danil Kaskov Vitaly Sychov Deni~ Avto PP 2009-~
#> 14 <NA> <NA> Vita~ Avto EV 2009-~
#> 15 <NA> <NA> Alex~ Avto PP 2009-~
#> 16 Artemy Panarin Stanislav Levin Pyot~ Russ~ EV 2009-~
所以，我想计算每个球员的进球数、主要助攻数和次要助攻数，然后为每个球员计算 1 行。假设“Artemy Panarin”这个名字被列为 1 次进球，0 次主要助攻，2 次次助攻，我的输出将如下所示：
tibble::tibble(name = c("Artemy Panarin", "Stanislav Levin", "Danil Kaskov"), team = c("Russkie Vityazi", "Russkie Vityazi", "Avto"), goals = c(1, 1, 0), primary_assists = c(0, 0, 1), secondary_assists = c(2, 0, 0))
#> # A tibble: 3 x 5
#> name team goals primary_assists secondary_assists
#> <chr> <chr> <dbl> <dbl> <dbl>
#> 1 Artemy Panarin Russkie Vityazi 1.00 0 2.00
#> 2 Stanislav Levin Russkie Vityazi 1.00 0 0
#> 3 Danil Kaskov Avto 0 1.00 0
这有意义吗？有任何想法吗？首选 Tidyverse 解决方案。谢谢！
【问题讨论】：
分组的逻辑不是很清楚 好的，我稍微编辑了我的问题。是不是更清楚了？ 我有一些预感，但到目前为止你有什么尝试？ 感谢@camille 的回复。由于某种原因，我只是无法理解这样做的逻辑，因为我习惯于直接进入mydata %>% group_by(name) %>% summarize(n_goals = count(goals), ...)。这就是我所了解的情况
类似gather(mydata, key, name, primary_assist:goal) %>% group_by(name, team, key) %>% summarise(n = n()) %>% spread(key, n, fill = 0)
【参考方案1】：
我们可以将gather 转换为'long' 格式，按'name'、'team' 和'key' 列分组（来自gather）、summarise 获取计数和spread 返回“宽”格式
library(tidyverse)
gather(mydata, key, name, primary_assist:goal) %>%
group_by(name, team, key) %>%
summarise(n = n()) %>%
spread(key, n, fill = 0)
# A tibble: 30 x 5
# Groups: name, team [30]
# name team goal primary_assist secondary_assist
# <chr> <chr> <dbl> <dbl> <dbl>
# 1 Alexander Antropov Avto 3 0 0
# 2 Alexander Streltsov Avto 1 0 0
# 3 Andrei Badrutdinov Belye Medvedi 0 0 1
# 4 Anton Burdasov Belye Medvedi 1 1 0
# 5 Artemy Panarin Russkie Vityazi 0 1 0
# 6 Bogdan Potekhin Stalnye Lisy 0 2 0
# 7 Daniil Apalkov Stalnye Lisy 1 0 1
# 8 Danil Gubarev Belye Medvedi 0 0 1
# 9 Danil Kaskov Avto 0 1 0
#10 Denis Belonogov Avto 1 0 0
# ... with 20 more rows
【讨论】：
@Frank 是的，我在想那个方向，但我想如果 OP 还想获得一些其他信息，那么summarise 会很有用【参考方案2】：
获得结果的一种方法是在汇总策略之外使用gather()/spread() 重塑数据。
library(tidyverse)
scoring_summary <- mydata %>%
select(primary_assist:team) %>%
gather("key", "player", -team) %>%
group_by(player) %>%
count(key) %>%
spread(key, n)
# convert NAs to 0
scoring_summary[is.na(scoring_summary)] <- 0
scoring_summary
# A tibble: 28 x 4
# Groups: player [28]
player goal primary_assist secondary_assist
<chr> <dbl> <dbl> <dbl>
1 Alexander Antropov 3 0 0
2 Alexander Streltsov 1 0 0
3 Andrei Badrutdinov 0 0 1
4 Anton Burdasov 1 1 0
5 Artemy Panarin 0 1 0
6 Bogdan Potekhin 0 2 0
7 Daniil Apalkov 1 0 1
8 Danil Gubarev 0 0 1
9 Danil Kaskov 0 1 0
10 Denis Belonogov 1 0 0
count() 与您最初尝试使用 summarise(count(goals) 时所做的相同
【讨论】：
你不需要在 count 前面加上 group_by，而 count(player, key) 似乎也一样。【参考方案3】：您可以使用收集和传播。先将进球和助攻列聚集成一个“键”，然后按键和球员分组。您可以稍后将 NA 转换为 0
library(tidyverse)
mydata_tidy <- mydata %>%
gather(key = "key", value = "player", primary_assist, secondary_assist, goal) %>%
na.omit()
mydata_tidy %>%
group_by(key, player) %>%
summarize(count = n()) %>%
spread(key, count) %>%
filter(player %in% c("Artemy Panarin", "Stanislav Levin", "Danil Kaskov"))
#> # A tibble: 3 x 4
#> player goal primary_assist secondary_assist
#> <chr> <int> <int> <int>
#> 1 Artemy Panarin NA 1 NA
#> 2 Danil Kaskov NA 1 NA
#> 3 Stanislav Levin NA NA 1
由reprex package (v0.2.0) 于 2018 年 7 月 18 日创建。
【讨论】：
以上是关于按名称汇总跨多个变量的数据的主要内容，如果未能解决你的问题，请参考以下文章