我在R'HLS'中有一个数据框，当访问者访问网站时，它基本上是页面明细的细节。如果他已经移动到page_count表示的第10页，则每行代表从第3页开始到最多10页的每页访问。
ID page_count purchase_flag prob hl_flag
V1 3 1 0.76 1
V1 4 1 0.65 1
V1 5 1 0.04 0
V1 6 1 0.86 1
V1 7 1 0.04 0
V1 8 1 0.65 1
V1 9 1 0.01 0
V1 10 1 0.00 0
V2 3 0 0.03 0
V2 4 0 0.01 0
V2 5 0 0.02 0
V2 6 0 0.00 0
V3 3 1 0.02 0
V3 4 1 0.001 0
V3 5 1 0.76 1
V3 6 1 0.03 0
V4 3 0 0.04 0
V4 4 0 0.65 1
V4 5 0 0.03 0
我想创建一个表，它将获取行直到第一次出现hl_flag = 1（如果该情况为真）并且如果hl_flag = 0则所有行都为任何ID。输出需要看起来像这样
ID page_count purchase_flag prob hl_flag
V1 3 1 0.76 1
V2 3 0 0.03 0
V2 4 0 0.01 0
V2 5 0 0.02 0
V2 6 0 0.00 0
V3 3 1 0.02 0
V3 4 1 0.001 0
V3 5 1 0.76 1
V4 3 0 0.04 0
V4 4 0 0.65 1
提前感谢您的帮助。
更新： 添加dput的输出如下
structure(list(ung_id = c("00000f23-1019-4aff-8199-35bd0d032356/1",
"00000f23-1019-4aff-8199-35bd0d032356/1", "00000f23-1019-4aff-8199-35bd0d032356/1",
"00000f23-1019-4aff-8199-35bd0d032356/1", "00002b20-82d4-497b-a137-34e3bb4eaf74/1",
"00002b20-82d4-497b-a137-34e3bb4eaf74/1", "00002b20-82d4-497b-a137-34e3bb4eaf74/1",
"0000aeff-2d8b-4daa-a084-fb2980f1feed/1", "0000aeff-2d8b-4daa-a084-fb2980f1feed/1",
"0000b96e-566f-4b6e-925a-b7dcfd4a7208/1", "0000b96e-566f-4b6e-925a-b7dcfd4a7208/1",
"0000b96e-566f-4b6e-925a-b7dcfd4a7208/1", "0000b96e-566f-4b6e-925a-b7dcfd4a7208/1",
"0000b96e-566f-4b6e-925a-b7dcfd4a7208/1", "0000b96e-566f-4b6e-925a-b7dcfd4a7208/1",
"0000b96e-566f-4b6e-925a-b7dcfd4a7208/1", "0000b96e-566f-4b6e-925a-b7dcfd4a7208/1",
"0000d089-edda-4c8b-8b17-d9def3cae7cf/1", "0000d089-edda-4c8b-8b17-d9def3cae7cf/1",
"0000d089-edda-4c8b-8b17-d9def3cae7cf/1"), nop_count = c(3L,
4L, 5L, 6L, 3L, 4L, 5L, 3L, 4L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L,
3L, 4L, 5L), purchase_flag = c(1L, 1L, 1L, 1L, 0L, 0L, 0L, 0L,
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L), prob = c(0.0777615841278747,
0.0738346887497272, 0.0741130887754292, 0.0785370078084892, 0.0619573259953132,
0.0516201527986966, 0.0562025814090338, 0.0837301511694211, 0.0579033581198143,
0.0364358545936557, 0.0329682922619259, 0.0420157964561273, 0.049855260762479,
0.0500481302257314, 0.0463893143028813, 0.049855260762479, 0.0391886960037603,
0.0683568422952682, 0.0570168506417919, 0.0661965354597502),
decile = structure(c(8L, 8L, 8L, 8L, 6L, 4L, 5L, 8L, 5L,
1L, 1L, 2L, 4L, 4L, 3L, 4L, 2L, 7L, 5L, 7L), .Label = c("(0.0257,0.0364]",
"(0.0364,0.0428]", "(0.0428,0.0482]", "(0.0482,0.0531]",
"(0.0531,0.0583]", "(0.0583,0.0645]", "(0.0645,0.0722]",
"(0.0722,0.0842]"), class = "factor"), hl_Flag = c(1L, 1L,
1L, 1L, 1L, 0L, 1L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L,
1L, 1L, 1L)), .Names = c("ung_id", "nop_count", "purchase_flag",
"prob", "decile", "hl_Flag"), row.names = c(NA, -20L), .internal.selfref = <pointer: 0x00000000002b0788>, class = c("data.table",
"data.frame"))
答案 0 :(得分：2)
一个选项是使用data.table。我们将'data.frame'转换为'data.table'（setDT(HLS)），按'ID'分组，我们检查是否有any'hl_flag'值为1.在这种情况下，我们使用which.max得到hl_flag中第一次出现1的索引，得到序列（1:(which.max..)），找到行索引（.I）或else只返回行索引（.I），提取具有行索引（$V1）的列，并使用它来对行进行子集化。
library(data.table)
setDT(HLS)[HLS[, if(any(hl_flag==1)) .I[1:(which.max(hl_flag))]
else .I, ID]$V1]
# ID page_count purchase_flag prob hl_flag
# 1: V1 3 1 0.760 1
# 2: V2 3 0 0.030 0
# 3: V2 4 0 0.010 0
# 4: V2 5 0 0.020 0
# 5: V2 6 0 0.000 0
# 6: V3 3 1 0.020 0
# 7: V3 4 1 0.001 0
# 8: V3 5 1 0.760 1
# 9: V4 3 0 0.040 0
#10: V4 4 0 0.650 1
或类似于I showed的方法data.table，base R选项将
do.call(rbind, lapply(split(HLS, HLS$ID),
function(x) if(any(x$hl_flag==1))
x[seq(which.max(x$hl_flag)), ]
else x))
或使用dplyr
library(dplyr)
HLS %>%
group_by(ID) %>%
filter(all(!hl_flag)| row_number() %in% seq(which.max(hl_flag)))
# ID page_count purchase_flag prob hl_flag
# (chr) (int) (int) (dbl) (int)
#1 V1 3 1 0.760 1
#2 V2 3 0 0.030 0
#3 V2 4 0 0.010 0
#4 V2 5 0 0.020 0
#5 V2 6 0 0.000 0
#6 V3 3 1 0.020 0
#7 V3 4 1 0.001 0
#8 V3 5 1 0.760 1
#9 V4 3 0 0.040 0
#10 V4 4 0 0.650 1
答案 1 :(得分：0)
你可以尝试，
l <- lapply(split(df, df$ID), function(x) {if(any(x[5] == 1)) x[1:which.max(x[5] == 1),] else x})
将df拆分为id，然后将每个列表分组到任何hl_flag == 1
哪个会给你一个明智的名单
#$V1
# ID page_count purchase_flag prob hl_flag
#1 V1 3 1 0.76 1
#$V2
# ID page_count purchase_flag prob hl_flag
#9 V2 3 0 0.03 0
#10 V2 4 0 0.01 0
#11 V2 5 0 0.02 0
#12 V2 6 0 0.00 0
#$V3
# ID page_count purchase_flag prob hl_flag
#13 V3 3 1 0.020 0
#14 V3 4 1 0.001 0
#15 V3 5 1 0.760 1
#$V4
# ID page_count purchase_flag prob hl_flag
#17 V4 3 0 0.04 0
#18 V4 4 0 0.65 1
要获得预期结果，您可以使用bind_rows包
dplyr
library(dplyr)
bind_rows(l)
#ID page_count purchase_flag prob hl_flag
#(fctr) (int) (int) (dbl) (int)
#1 V1 3 1 0.760 1
#2 V2 3 0 0.030 0
#3 V2 4 0 0.010 0
#4 V2 5 0 0.020 0
#5 V2 6 0 0.000 0
#6 V3 3 1 0.020 0
#7 V3 4 1 0.001 0
#8 V3 5 1 0.760 1
#9 V4 3 0 0.040 0
#10 V4 4 0 0.650 1