提前感谢您对此有任何帮助或建议。这是我正在使用的数据帧的缩写示例。
boxscore_stats = structure(list(game_id = c(157046L, 157046L, 157046L, 157046L,
157046L, 157046L, 157046L, 157046L, 157046L, 157046L, 157046L,
157046L, 157046L, 157046L, 157046L, 157046L, 157046L, 157046L,
159151L, 159151L, 159151L, 159151L, 159151L, 159151L, 159151L,
159151L, 159151L, 159151L, 159151L, 159151L, 159151L, 159151L,
159151L, 159151L, 159151L, 159151L, 159151L, 159151L, 159151L,
159151L), team_id = c(116975, 116975, 116975, 116975, 116975,
116975, 116975, 116975, 116975, 120310, 120310, 120310, 120310,
120310, 120310, 120310, 120310, 120310, 121910, 121910, 121910,
121910, 121910, 121910, 121910, 121910, 121910, 121910, 122072,
122072, 122072, 122072, 122072, 122072, 122072, 122072, 122072,
122072, 122072, 122072), minutes_played = c(18.76, 14.63, 8,
16.69, 24.62, 32, 12.79, 5.28, 3.22, 24.35, 10.18, 20.65, 9.59,
25.08, 14.12, 17.46, 23.15, 15.43, 22.84, 19.27, 21.31, 6.41,
17.57, 17.4, 17.29, 7.22, 12.09, 17.25, 2.28, 16.87, 6.6, 19.73,
6.31, 13.25, 26.25, 6.08, 28.71, 11.2, 17.54, 5.17), fieldGoalsMade = c(1L,
1L, 4L, 1L, 2L, 7L, 1L, 1L, 1L, 4L, 0L, 3L, 1L, 3L, 0L, 6L, 7L,
1L, 7L, 4L, 5L, 1L, 2L, 6L, 2L, 0L, 1L, 3L, 0L, 1L, 1L, 3L, 0L,
1L, 11L, 2L, 5L, 1L, 2L, 1L), fieldGoalAttempts = c(8L, 6L, 7L,
2L, 9L, 16L, 3L, 1L, 2L, 12L, 4L, 12L, 3L, 11L, 4L, 9L, 13L,
6L, 12L, 10L, 14L, 2L, 6L, 11L, 6L, 2L, 2L, 6L, 0L, 5L, 3L, 10L,
2L, 3L, 21L, 3L, 17L, 4L, 9L, 2L)), .Names = c("game_id", "team_id",
"minutes_played", "fieldGoalsMade", "fieldGoalAttempts"), row.names = c(NA,
40L), class = "data.frame")
head(boxscore_stats)
game_id team_id minutes_played fieldGoalsMade fieldGoalAttempts
1 157046 116975 18.76 1 8
2 157046 116975 14.63 1 6
3 157046 116975 8.00 4 7
4 157046 116975 16.69 1 2
5 157046 116975 24.62 2 9
6 157046 116975 32.00 7 16
7 157046 116975 12.79 1 3
8 157046 116975 5.28 1 1
9 157046 116975 3.22 1 2
10 157046 120310 24.35 4 12
11 157046 120310 10.18 0 4
12 157046 120310 20.65 3 12
13 157046 120310 9.59 1 3
14 157046 120310 25.08 3 11
15 157046 120310 14.12 0 4
16 157046 120310 17.46 6 9
17 157046 120310 23.15 7 13
18 157046 120310 15.43 1 6
19 159151 121910 22.84 7 12
20 159151 121910 19.27 4 10
21 159151 121910 21.31 5 14
22 159151 121910 6.41 1 2
23 159151 121910 17.57 2 6
24 159151 121910 17.40 6 11
25 159151 121910 17.29 2 6
26 159151 121910 7.22 0 2
27 159151 121910 12.09 1 2
28 159151 121910 17.25 3 6
29 159151 122072 2.28 0 0
30 159151 122072 16.87 1 5
31 159151 122072 6.60 1 3
32 159151 122072 19.73 3 10
33 159151 122072 6.31 0 2
34 159151 122072 13.25 1 3
35 159151 122072 26.25 11 21
36 159151 122072 6.08 2 3
37 159151 122072 28.71 5 17
38 159151 122072 11.20 1 4
39 159151 122072 17.54 2 9
40 159151 122072 5.17 1 2
关于这个数据框的重要注意事项是每个game_id对应两个team_id，对于在游戏中玩的两个团队。每个game_id对于一场篮球比赛来说都是独一无二的。每行对应于该游戏中team_ids团队中玩家的统计数据。上面的示例只有两个游戏/ 4个团队/ 40个玩家，但我的完整数据框有数百个游戏，每个团队都会出现多次。
我能够做的第一个聚合是通过team_id聚合所有内容。这段代码为我完成了第一次聚合的工作：
boxscore_stats_aggregated = aggregate(boxscore_stats, by = list(boxscore_stats[, 2]), FUN = sum)
这是相当简单的。对于任何team_id，我已经汇总了他们所有的分钟，他们所有的fieldGoalsMade等等。对于我的下一个聚合，我需要再次通过team_id聚合，而不是通过他们自己的行/统计数据聚合团队，而是我需要汇总对手的行/统计数据。这回答了这个问题＆＃34;对于任何一支球队来说，他们总共允许多少场比赛马匹给对手等等。＆＃34;所以在这种情况下，对于team_id = 116975，我想用team_id 120310聚合所有行。当然，下一次team_id 116975出现在我的数据帧中的新游戏中，很可能他们正在扮演一个不同的对手，所以这个聚合并不像team_id 120310聚合那么简单。
我认为我应该能够使用两个team_id之间的关系，这对于独特的game_ids是唯一的，以使这种聚合成为可能，但我正在努力实现它的实现方式。
谢谢！
答案 0 :(得分：2)
这是使用data.table的方法：
（1）读入数据：
# Load package
library(data.table)
# Load your data
boxscore_stats <- fread("row game_id team_id minutes_played fieldGoalsMade fieldGoalAttempts
1 157046 116975 18.76 1 8
2 157046 116975 14.63 1 6
3 157046 116975 8.00 4 7
4 157046 116975 16.69 1 2
5 157046 116975 24.62 2 9
6 157046 116975 32.00 7 16
7 157046 116975 12.79 1 3
8 157046 116975 5.28 1 1
9 157046 116975 3.22 1 2
10 157046 120310 24.35 4 12
11 157046 120310 10.18 0 4
12 157046 120310 20.65 3 12
13 157046 120310 9.59 1 3
14 157046 120310 25.08 3 11
15 157046 120310 14.12 0 4
16 157046 120310 17.46 6 9
17 157046 120310 23.15 7 13
18 157046 120310 15.43 1 6
19 159151 121910 22.84 7 12
20 159151 121910 19.27 4 10
21 159151 121910 21.31 5 14
22 159151 121910 6.41 1 2
23 159151 121910 17.57 2 6
24 159151 121910 17.40 6 11
25 159151 121910 17.29 2 6
26 159151 121910 7.22 0 2
27 159151 121910 12.09 1 2
28 159151 121910 17.25 3 6
29 159151 122072 2.28 0 0
30 159151 122072 16.87 1 5
31 159151 122072 6.60 1 3
32 159151 122072 19.73 3 10
33 159151 122072 6.31 0 2
34 159151 122072 13.25 1 3
35 159151 122072 26.25 11 21
36 159151 122072 6.08 2 3
37 159151 122072 28.71 5 17
38 159151 122072 11.20 1 4
39 159151 122072 17.54 2 9
40 159151 122072 5.17 1 2
")
（2）继续进行实际计算：
# Aggregate on team-and game level (data.table style)
boxscore_stats_aggregated <- boxscore_stats[, lapply(.SD, sum), by = list(game_id, team_id)]
# Match EVERY team to each opponent, i.e. still two rows per game
# but columns for opponent's performance added.
# Some teams drops out in the dummy data as they opponent data was missing.
merge(boxscore_stats_aggregated, boxscore_stats_aggregated,
by="game_id", suffixes = c("", ".opponent"))[team_id!=team_id.opponent,]
output看起来像这样：
# > output
# game_id team_id row minutes_played fieldGoalsMade fieldGoalAttempts team_id.opponent row.opponent minutes_played.opponent fieldGoalsMade.opponent fieldGoalAttempts.opponent
# 1: 1413414 116975 45 135.99 19 54 120310 126 160.01 25 74
# 2: 1413414 120310 126 160.01 25 74 116975 45 135.99 19 54
答案 1 :(得分：1)
以防万一，OP考虑或未来的读者是merge()的基础R版本，用于game_id的团队和反对的并排聚合。需要一个临时临时文件gamecount。
# TEAM AGGREGATION
aggdf <- aggregate(.~game_id + team_id, boxscore_stats, FUN = sum)
# GAME COUNT BY TEAM (TEMP COL USED FOR MERGE/FILTER)
aggdf$gamecount <- sapply(1:nrow(aggdf), function(i)
sum(aggdf[1:i, c("game_id")] == aggdf$game_id[i]))
# MERGE AND FILTER
mdf <- merge(aggdf, aggdf, by="game_id")
mdf <- mdf[mdf$team_id.x != mdf$team_id.y & mdf$gamecount.x == 1,]
mdf$gamecount.x <- mdf$gamecount.y <- NULL
# RENAME COL AND ROW NAMES
names(mdf)[grepl("\\.x", names(mdf))] <- gsub("\\.x", "",
names(mdf)[grepl("\\.x", names(mdf))])
names(mdf)[grepl("\\.y", names(mdf))] <- gsub("\\.y", ".opp",
names(mdf)[grepl("\\.y", names(mdf))])
rownames(mdf) <- 1:nrow(mdf)
# game_id team_id minutes_played fieldGoalsMade fieldGoalAttempts team_id.opp
# 1 157046 116975 135.99 19 54 120310
# 2 159151 121910 158.65 31 71 122072
# minutes_played.opp fieldGoalsMade.opp fieldGoalAttempts.opp
# 1 160.01 25 74
# 2 159.99 28 79
答案 2 :(得分：0)
如果你想隔离单个team_id，我会使用dplyr包。
例如，如果您想知道每个团队的字段目标百分比，我会写出类似的内容：
LinearLayout.LayoutParams l=new LinearLayout.LayoutParams();
l.width = l.height = LinearLayout.LayoutParams.WRAP_CONTENT;
这会为您提供由团队boxscore_stats %>%
group_by(team_id) %>%
summarize(perc_fg = sum(fieldGoalsMade)/sum(fieldGoalAttempts))
汇总的新data.frame。