如何获取自定义函数并将其与 r 中的 dplyr group_by 一起使用

很抱歉问了同样的问题，但我就是不知道如何使用自定义函数并在分组数据上迭代它.所以，如果有人能给我指出一些深入的资源，我会得到额外的分数.

这样做的目的是:

library(dplyr)

a_df <- tibble(team = "A",
             cluster = list(
               c(0.01, 0.01, 0.09, 0.03, 0.14, 0.28, 0.09, 0.25, 0.18, 0.17, 0.54, 0.41, 0.16, 0.18, 0.25, 0.02, 0.2, 0.69, 0.02, 0.01, 0.02, 0.07, 0.07, 0.21)),
             tot_matches = 20,
             tot_ck = 121)

#Function that will take a list of xG and return the probability of n goals being scored.Normalized
calculate_goals_norm <- function(shot_xg_list, n_matches, norm_matches){
  
  #set seed only for stackoverflow question!!!!
  set.seed(123)
  
  #Start goal total at 0
  goals_total <- 0
  
  #Function to take an xG number and simulate whether it was a goal or not
  xg_to_goals_sim <- function(shot_xg){

    #Start goal count at 0
    Goals <- 0

    #For each shot, if it goes in, add a goal
    for (shot in shot_xg){
      if (runif(1)<=shot){
        Goals <- Goals + 1
      }
    }

    #Finally, return the number of goals
    return(Goals)

  }
  
  #Run xG calculator 10000 times to test winner %
  sim_goal_list <- c()
  # Create a for statement to populate the list
  for (i in 1:10000) {
    #Run the above formula for xG list
    #But first we need to normalize the number of xG by randomly picking from xG
    
    if(n_matches == norm_matches){
      new_shot_xg_list <- shot_xg_list
      goals_total <- xg_to_goals_sim(new_shot_xg_list)
      sim_goal_list[[i]] <- goals_total
    }
    
    if(n_matches < norm_matches){
      new_shot_xg_list <- c(shot_xg_list, 
                            sample(shot_xg_list, 
                                   round(length(shot_xg_list)/n_matches*norm_matches,0)-length(shot_xg_list),
                                   replace=FALSE))
      goals_total <- xg_to_goals_sim(new_shot_xg_list)
      sim_goal_list[[i]] <- goals_total
    }
    
    if(n_matches > norm_matches){
      
      new_shot_xg_list <- sample(shot_xg_list, 
                                 round(length(shot_xg_list)/n_matches*norm_matches,0),
                                 replace=FALSE)
      goals_total <- xg_to_goals_sim(new_shot_xg_list)
      sim_goal_list[[i]] <- goals_total
    }
    
  }
  
  
  sim_goal_dat <- data.frame(value = unlist(sim_goal_list))
  
  goal_prob <- sim_goal_dat %>% 
    count(value) %>% 
    summarise(goals = value,
              prob = round(n/1000*10,1)) %>% 
    arrange(goals)
  
  return(goal_prob)
  
  
}

#apply function to a single team dataframe (1 obs. of 4 variables)
calculate_goals_norm(shot_xg_list= unlist(a_df$cluster), n_matches = a_df$tot_matches, norm_matches = 20)

# A tibble: 12 x 2
   goals  prob
   <dbl> <dbl>
 1     0   0.6
 2     1   4  
 3     2  12.4
 4     3  20.6
 5     4  23.8
 6     5  19.1
 7     6  11.6
 8     7   5.6
 9     8   1.8
10     9   0.4
11    10   0.1
12    12   0 

添加第二个团队来创建一个full_df，以实现定制功能的预期用途:

full_df <- add_row(a_df, tibble(team = "B",
                                cluster = list(
                                  c(0.06, 0.01, 0.11, 0.18, 0.75, 0.04, 0.23, 0.07, 0.1, 0.05, 0.24, 0.12, 0.28, 0.02, 0.09, 0.16, 0.64, 0.03, 0.1, 0.19, 0.09, 0.01, 0.02, 0.12, 0.01, 0.11, 0.18, 0.05, 0.02, 0.8, 0.08)),
                                tot_matches = 19,
                                tot_ck = 83) )

#final function should look like this?
full_df %>% 
  group_by(team) %>% 
  calculate_goals_norm(shot_xg_list = unlist(cluster), n_matches = tot_matches, norm_matches = 20)

我不喜欢上面例子中的自定义函数.我知道apply()/sapply()通常用于在数据帧上进行迭代，但我还是没有足够的知识来知道如何在这里apply.

谢谢你的帮助.