在不同的数据范围内提取和减去行r的更有效的方法

Question

我正在使用该篮球游戏数据，其中包含约50,000行的数据帧游戏。我正在尝试比较每个游戏中每个团队（A和B）的统计数据。

我还有另一个名为TeamStats的数据框架，每个赛季都有大约3000行。

到目前为止，我已经组装了一个代码：

    for (i in 1:nrow(games)) {
  if (length(which(((teamStats$Year == games$Season[i])==1) & (teamStats$teamID == games$teamA[i]))) == 1) {
    selectTeamA <- teamStats[which(((teamStats$Year == games$Season[i])==1) & (teamStats$teamID == games$teamA[i])),4:45]
  } else {
    selectTeamA <- as.numeric(rep(NA, ncol(differences)))
  }
  if (length(which(((teamStats$Year == games$Season[i])==1) & (teamStats$teamID == games$teamB[i]))) == 1) {
    selectTeamB <- teamStats[which(((teamStats$Year == games$Season[i])==1) & (teamStats$teamID == games$teamB[i])),4:45]
  } else {
    selectTeamB <- as.numeric(rep(NA, ncol(differences)))
  }
  differences[i,] <- selectTeamA - selectTeamB
}

基本上，此代码在征服了正确的赛季后，为每个团队A和B搜索了正确的团队ID。由于TeamStats中没有每个赛季的每个球队，因此我现在已经用NA填补了失踪的行。"差异"数据框是一个空的数据框，它将填补我的for循环中A和B统计数据的差异。

让您了解数据：

游戏 - 前6行

           Season teamA teamB winner scoreA scoreB
108123   2010  1143  1293      A     75     70
108124   2010  1198  1314      B     72     88
108125   2010  1108  1326      B     60    100
108126   2010  1107  1393      B     43     75
108127   2010  1143  1178      A     95     61

teamStats-前6行，仅用于空间的前6列 - 许多列在完整数据帧中的列表。该代码为TeamID找到正确的行，然后取得统计列，例如G w l etc

              School Year teamID  G  W  L
1  abilene christian 2018   1101 32 16 16
2          air force 2018   1102 31 12 19
3              akron 2018   1103 32 14 18
4        alabama a&m 2018   1105 31  3 28
5 alabama-birmingham 2018   1412 33 20 13

，要关闭这篇很长的帖子，我的问题。我的循环代码有效，并填充差异数据框。问题是运行此代码需要20-30分钟。我并不是很有经验来处理这么多数据。我不知道有一种技术吗？如何以更有效的方式重写此代码？

Answer 1

这是一种使用tidyverse软件包的方法，我期望它应该比OP中的循环解决方案快得多。速度（我希望）来自更多地依靠数据库加入操作（例如基本merge或DPLYR的left_join）来连接两个表。

library(tidyverse)
# First, use the first few columns from the `games` table, and convert to long format with
#   a row for each team, and a label column `team_cat` telling us if it's a teamA or teamB.
stat_differences <- games %>%
  select(row, Season, teamA, teamB)  %>% 
  gather(team_cat, teamID, teamA:teamB) %>%  
# Join to the teamStats table to bring in the team's total stats for that year
  left_join(teamStats %>% select(-row),    # We don't care about this "row"
            by = c("teamID", "Season" = "Year")) %>%
# Now I want to reverse the stats' sign if it's a teamB. To make this simpler, I gather
#   all the stats into long format so that we can do the reversal on all of them, and 
#   then spread back out.
  gather(stat, value, G:L) %>%
  mutate(value = if_else(team_cat == "teamB", value * -1, value * 1)) %>%
  spread(stat, value) %>%
# Get the difference in stats for each row in the original games table.
  group_by(row) %>%
  summarise_at(vars(G:W), sum)
# Finally, add the output to the original table
output <- games %>% 
  left_join(stat_differences)

为了测试此问题，我更改了给定的示例数据，以使两个表将彼此相关：

games <- read.table(header = T, stringsAsFactors = F,
  text = "row           Season teamA teamB winner scoreA scoreB
108123   2010  1143  1293      A     75     70
108124   2010  1198  1314      B     72     88
108125   2010  1108  1326      B     60    100")
teamStats <- read.table(header = T, stringsAsFactors = F,
  text = "row   School Year teamID  G  W  L
1  abilene_christian 2010   1143 32 16 16
2          air_force 2010   1293 31 12 19
3              akron 2010   1314 32 14 18
4        alabama_a&m 2010   1198 31  3 28
5 alabama-birmingham 2010   1108 33 20 13
6       made_up_team 2018   1326 160 150 10    # To confirm getting right season
7       made_up_team 2010   1326 60 50 10"
)

然后我得到以下输出，这似乎很有意义。（我刚刚意识到我应用的聚集/突变/传播更改了列的顺序；如果我有时间，我可能会尝试使用mutate_if保留顺序。）

> output
     row Season teamA teamB winner scoreA scoreB   G  L   W
1 108123   2010  1143  1293      A     75     70   1 -3   4
2 108124   2010  1198  1314      B     72     88  -1 10 -11
3 108125   2010  1108  1326      B     60    100 -27  3 -30

Answer 2

一种方法是合并 games和 teamStats，作为跨行迭代的替代方法。

一些代码复制您的设置，以创建一个最小的工作示例：

library(dplyr)
library(purrr)
set.seed(123)
n_games <- 50000
n_teams <- 400
n_years <- 10
games <- data.frame(Season = rep(2005:(2005 + n_years - 1),
                                 each = n_games / n_years)) %>%
  mutate(teamA = sample(1000:(1000 + n_teams - 1), n_games, r = TRUE),
         teamB = map_int(teamA, ~sample(setdiff(1000:(1000 + n_teams - 1), .), 1)),
         scoreA = as.integer(rnorm(n_games, 80, 20)),
         scoreB = as.integer(rnorm(n_games, 80, 20)),
         scoreB = ifelse(scoreA == scoreB, scoreA + sample(c(-1, 1), n_games, r = TRUE), scoreB),
         winner = ifelse(scoreA > scoreB, "A", "B"))
gen_random_string <- function(...) {
  paste(sample(c(letters, " "), rpois(1, 10), r = TRUE), collapse = "")
}
schools_ids <- data.frame(teamID = 1000:(1000 + n_teams - 1)) %>%
  mutate(School = map_chr(teamID, gen_random_string))
teamStats <- data.frame(Year = rep(2005:(2005 + n_years - 1), each = 300)) %>%
  mutate(teamID = as.vector(replicate(n_years, sample(schools_ids$teamID, 300))),
         G = 32, W = rpois(length(teamID), 16), L = G - W) %>%
  left_join(schools_ids)

我们的games具有50k行和3K行的TeamStats。现在，我们通过Year和teamID将teamStats倒入tibble：

teamStats <- teamStats %>%
  group_by(Year, teamID) %>%
  nest()
# # A tibble: 3,000 x 3
#     Year teamID data            
#    <int>  <int> <list>          
#  1  2005   1321 <tibble [1 x 4]>
#  2  2005   1192 <tibble [1 x 4]>
#  3  2005   1074 <tibble [1 x 4]>
# <snip>

做出较小的便利功能来计算差异：

calculate_diff <- function(x, y) {
  if (is.null(x) | is.null(y)) {
    data.frame(G = NA, W = NA, L = NA)
  } else {
    x[, 1:3] - y[, 1:3]
  }
}

现在，我们（1）与teamStats加入（或合并）games，（2）使用加入的数据集计算差异，（3）unnest（或Un-collapse）dataFrame。

。

start <- Sys.time()
differences <- games %>%
  left_join(teamStats, c("Season" = "Year", "teamA" = "teamID")) %>%
  rename(teamA_stats = data) %>%
  left_join(teamStats, c("Season" = "Year", "teamB" = "teamID")) %>%
  rename(teamB_stats = data) %>%
  mutate(diff = map2(teamA_stats, teamB_stats, calculate_diff)) %>%
  select(Season, teamA, teamB, diff) %>%
  unnest()
difftime(Sys.time(), start)
# Time difference of 11.27832 secs

结果

head(differences)
#   Season teamA teamB  G  W  L
# 1   2005  1115  1085 NA NA NA
# 2   2005  1315  1177 NA NA NA
# 3   2005  1163  1051  0 -9  9
# 4   2005  1353  1190  0 -4  4
# 5   2005  1376  1286 NA NA NA
# 6   2005  1018  1362  0 -1  1