我有以下df（我的实际df更大，但这更好解释）
df
Date Team Result
1 20-10-2014 Bulls W
2 20-11-2014 Bulls W
3 21-11-2014 Bulls W
4 22-10-2014 Bulls L
5 23-10-2014 Bulls L
6 24-10-2014 Bulls W
7 25-10-2014 Bulls W
8 26-10-2014 Bulls L
我用rle来计算连胜条件
df$Result <- as.character(df$Result)
result.rle <- rle(df$Result)
最大条纹数量
sort(w.rle$lengths, decreasing = TRUE)
我现在要做的是用2以上的连胜来过滤相关的比赛。
我检查了?rle，但无法找到我要找的东西。有关如何做到这一点的任何想法？
我的最终结果应该是：
#winning streak > 2
Date Team Result
1 20-10-2014 Bulls W
2 20-11-2014 Bulls W
3 21-11-2014 Bulls W
答案 0 :(得分：4)
这是使用data.table::rleid函数
library(data.table) # V 1.9.6+
setDT(df)[, indx := .N, by = rleid(Result)][Result == "W" & indx > 2]
# Date Team Result indx
# 1: 20-10-2014 Bulls W 3
# 2: 20-11-2014 Bulls W 3
# 3: 21-11-2014 Bulls W 3
另一种选择是破解by声明，以便我们可以在同一个调用中重复使用Result列（而无需单独创建indx）
df[, .SD[.N > 2 & Result == "W"], by = rleid(df[["Result"]])]
# rleid Date Team Result
# 1: 1 20-10-2014 Bulls W
# 2: 1 20-11-2014 Bulls W
# 3: 1 21-11-2014 Bulls W
答案 1 :(得分：1)
使用df[intersect(which(rep(rle(df$Result)$lengths,rle(df$Result)$lengths)>2),which(rep(rle(df$Result)$values,rle(df$Result)$lengths)=='W')),]
# Date Team Result
# 1 20-10-2014 Bulls W
# 2 20-11-2014 Bulls W
# 3 21-11-2014 Bulls W
的解决方案：
df <- structure(list(Date = c("20-10-2014", "20-11-2014", "21-11-2014",
"22-10-2014", "23-10-2014", "24-10-2014", "25-10-2014", "26-10-2014"
), Team = c("Bulls", "Bulls", "Bulls", "Bulls", "Bulls", "Bulls",
"Bulls", "Bulls"), Result = c("W", "W", "W", "L", "L", "W", "W",
"L")), .Names = c("Date", "Team", "Result"), class = "data.frame", row.names = c(NA,
-8L))
数据：
rle(df$Result) rle(df$Result)$lengths返回两个元素的列表：重复次数和重复值。我们复制长度（我们可以使用intersect）来获得与df行数相同的向量。然后我们选择具有重复＆gt;的行。 2。
which(rep(rle(df$Result)$values,rle(df$Result)$lengths)=='W')
# [1] 1 2 3 6 7 # Those are the rows where it's a Win. Only those rows can be included
which(rep(rle(df$Result)$lengths,rle(df$Result)$lengths)>2)
# [1] 1 2 3 # those are the rows where it's a streak of 3 wins or losses
取行的索引的交集：
详细说明：
H2S 1.4
C1 3.6
交叉点将给出预期的输出。