从 R 中的数据点提取一些数字的最有效方法是什么？ （加上其他具体步骤！）

Question

I've got quite a specific problem, for which I can just about find a very hacky solution, but I'm hoping somebody could outline a slightly more elegant method.

I have a CSV file, consisting of one row per historical football match played. The fields I care about look something like this:

home_team <- c("Team A", "Team B", "Team B")
away_team <- c("Team C", "Team C", "Team D")
home_goals <- c(2, 0, 1)
away_goals <- c(1, 2, 0)
home_goal_mins <- c("5 60", "NA", "80")
away_goal_mins <- c("15", "20 40", "NA")

df <- data.frame(home_team, away_team, home_goals, away_goals, home_goal_mins, away_goal_mins,
                 stringsAsFactors = FALSE)

df
#>   home_team away_team home_goals away_goals home_goal_mins away_goal_mins
#> 1    Team A    Team C          2          1           5 60             15
#> 2    Team B    Team C          0          2             NA          20 40
#> 3    Team B    Team D          1          0             80             NA

^{Created on 2020-10-05 by the reprex package (v0.3.0)}

My goal is to transform this dataframe such that there is one line per goal scored, per game, like this:

The main challenges, as I see them:

1. The *_goal_mins fields are read in as strings containing both numbers and NAs
2. Replicating the rows such that the Home/Away team combinations have the same number of rows as the total number of goals for that match

With regards to (1), I've been using stringr::str_split(., " ") to extract the numbers but then struggle to transform them into a numeric vector. Taking the first row of df as an example, I'm struggling to transform "5 60" into c(5, 60) , and it gets harder for me when I try to combine the home team's "5 60" with the away team's "15" to get the full goal sequence of c(5, 15, 60) .

As for (2), my current approach is to calculate the total_goals_scored per match, and do the following:

expanded_df <- df[rep(seq_len(dim(df)[1]),
                      df$total_goals_scored), ]

but I sense that there may be a better method.

Any help or tips will be appreciated! Thanks

Answer 1

Using dplyr and tidyr library you could do

1. bring home_goal_mins and away_goal_mins in same column using pivot_longer .
2. Split the data on whitespace and separate the goals in separate rows
3. Drop NA values
4. arrange data based on timestamp
5. Get data in wide format.

library(dplyr)
library(tidyr)

df %>%
  pivot_longer(cols = c(home_goal_mins, away_goal_mins)) %>%
  separate_rows(value, sep = ' ', convert = TRUE) %>%
  filter(!is.na(value)) %>%
  arrange(home_team, away_team, value) %>%
  group_by(home_team, away_team) %>%
  mutate(row = row_number()) %>%
  pivot_wider()

#  home_team away_team home_goals away_goals   row home_goal_mins away_goal_mins
#  <chr>     <chr>          <dbl>      <dbl> <int>          <int>          <int>
#1 Team A    Team C             2          1     1              5             NA
#2 Team A    Team C             2          1     2             NA             15
#3 Team A    Team C             2          1     3             60             NA
#4 Team B    Team C             0          2     1             NA             20
#5 Team B    Team C             0          2     2             NA             40
#6 Team B    Team D             1          0     1             80             NA