如何使用 R 评估分类变量随时间的变化？答案

【问题标题】：How does one evaluate changes in a categorical variable over time using R?如何使用 R 评估分类变量随时间的变化？
【发布时间】：2019-12-31 11:14:36
【问题描述】：

我有一个数据集，其中两支球队在一年一度的比赛中展开对决。这些比赛有两个分区，东部和西部。我想根据前一年比赛的结果来确定某一年的卫冕冠军是谁。我想为两个部门都这样做。

这是我的数据集：

data <- data.frame(
  Team = c("Hot Dogs", "Hamburgers", "Hot Dogs", "Hamburgers", "Hot Dogs",
           "Hamburgers", "Pho", "Ramen", "Pho", "Ramen", "Pho", "Ramen"),
  Division = c("West", "West", "West", "West", "West", "West", "East", "East",
               "East", "East", "East", "East"),
  Year = c("2017", "2017", "2018", "2018", "2019", "2019", "2017", "2017",
           "2018", "2018", "2019", "2019"),
  Score = c("37", "2", "26", "32", "37", "9", "22", "31", "25", "32", "24", "18"))

理想情况下，我会在原始数据中添加一个“结果”列，以指示给定球队是否是进入该比赛的卫冕冠军。像这样的：

data$Result <- c("Initial Champion", "NA", "Champion", "NA", "NA", "Champion", "NA", 
"Initial Champion", "NA", "Champion", "NA", "Champion")

是否有使用 R 的直接方法，特别是在可能的情况下使用 tidyverse 库？

感谢任何建议。提前致谢。

【问题讨论】：

抱歉，您能解释一下想要的结果df吗？
我想要原始数据框，但根据我上面描述的评估添加了结果列。我在最初的帖子中提供了结果列的样子。希望对您有所帮助。
检查dplyr.tidyverse.org/reference/lead-lag.html ...

标签： r

【解决方案1】：

首先我们得到一个包含所有冠军的表格，如果它是第一个，则将它们标记为“初始冠军”，将其他标记为“冠军”：

library(dplyr)
X = data %>% 
arrange(Year,desc(Score)) %>% 
group_by(Division) %>% 
filter(!duplicated(Year))%>% 
mutate(result=rep(c("Initial Champion","Champion"),times=c(1,n()-1)))

# A tibble: 6 x 5
# Groups:   Division [2]
  Team       Division Year  Score result          
  <fct>      <fct>    <fct> <fct> <chr>           
1 Hot Dogs   West     2017  37    Initial Champion
2 Ramen      East     2017  31    Initial Champion
3 Hamburgers West     2018  32    Champion        
4 Ramen      East     2018  32    Champion        
5 Hamburgers West     2019  9     Champion        
6 Pho        East     2019  24    Champion

要进入决赛桌，只需：

left_join(data,X)

【讨论】：

【解决方案2】：

在以下答案中，我使用 dplyr 来确定初始冠军和冠军，其中初始冠军意味着第一次出现数据的团队是其部门中年度最佳。在以后的几年里，在其部门中得分最高的球队被认为是冠军。

library(dplyr)

data <- data.frame(
  Team = c("Hot Dogs", "Hamburgers", "Hot Dogs", "Hamburgers", "Hot Dogs",
           "Hamburgers", "Pho", "Ramen", "Pho", "Ramen", "Pho", "Ramen"),
  Division = c("West", "West", "West", "West", "West", "West", "East", "East",
               "East", "East", "East", "East"),
  Year = c("2017", "2017", "2018", "2018", "2019", "2019", "2017", "2017",
           "2018", "2018", "2019", "2019"),
  Score = c("37", "2", "26", "32", "37", "9", "22", "31", "25", "32", "24", "18"),
  stringsAsFactors = FALSE)

result <-
  data %>%
  group_by(Year, Division) %>% # First we group by each year and division

  # For each division/year we get highest score then for the team with this score
  # we consider it champion
  mutate(high_score = as.character(max(as.numeric(Score), na.rm = TRUE)),
         result = ifelse(high_score == Score, "Champion", NA_character_)) %>%

  # Now to determine the initial champion we compare it with the first year
  # if the row contains data of the first year in data then it is initial
  mutate(result = 
           ifelse(min(data$Year) == Year & result == "Champion", "Initial Champion", result)) %>%

  # Here we drop high_score column because it is not needed in final output
  select(-high_score)

【讨论】：