使用重复值并在有断点时使用条件创建新列

Question

我的数据是通过遥测定位的约40只动物（id），我已经规定了3个区域。第一个是AR，这里是繁殖区域，AM迁移，AA是饲养区域。所有动物的第一个位置都在AR。但是有时动物还处于繁殖期（在AR），但是可以几次出到AM，然后又回到AR。只有当动物只有AM时，它们才开始迁移，直到到达饲养区域AA。因此，它们从AR开始，然后开始迁移AM，然后到达进料区域AA。

我正在尝试创建一些我不知道该怎么做的条件的新列，例如，我有此数据框

id     area   
2304   AR
2304   AR
2304   AR
2304   AM  #this AM for example, can repeat until 20 times and then came back to AR
2304   AM
2304   AR
2304   AR
2304   AR
2304   AM
2304   AM
2304   AM
2304   AM
2304   ...
2304   AM
2304   AM
2304   AM
2304   AA
2304   AA
2304   ...
2304   AA

因此，当有AR次x且此后有一个或直到20 AM并返回有AR时，我想要一个带有AR的新列。当有AM x次并且只有AM，而又没有回到AR时，我想要带有AM的新列。像这样：

AA没问题，AA =始终是AA

我期望这样：

id    area    fixed_area
2304   AR      AR
2304   AR      AR
2304   AR      AR
2304   AM      AR  #this AM for example, can repeat until 20 times and then came back to AR
2304   AM      AR
2304   AR      AR
2304   AR      AR
2304   AR      AR
2304   AM      AM
2304   AM      AM
2304   AM      AM
2304   AM      AM
2304   ...     ...
2304   AM      AM
2304   AM      AM
2304   AM      AM
2304   AA      AA
2304   AA      AA
2304   ...    ...
2304   AA      AA

我尝试过：

但是缺少AA，也许问题是因为需要对每只动物进行这种分离（id）

> table(df$area)

   AA    AM    AR 
31460 39101 28820 

class(df$area)
[1] "character"
> idx <- with(rle(as.character(df$area)), rep(seq_along(lengths), lengths))
> df$fixed_area <- with(df, replace(area, idx < max(idx[area == 'AM']), 'AR'))
> table(df$fixed_area)

   AM    AR 
  145 99236 
> 

此后，我放入数据帧，但是我的数据帧有90.000多行，所以我只复制了头值

> dput(head(df))
structure(list(DeployID = c("111868_16", "111868_16", "111868_16", 
"111868_16", "111868_16", "111868_16"), Start = structure(c(1477323868, 
1477323946, 1477324002, 1477324044, 1477324260, 1477324480), class = c("POSIXct", 
"POSIXt"), tzone = "GMT"), End = structure(c(1477323944, 1477324000, 
1477324042, 1477324170, 1477324458, 1477324542), class = c("POSIXct", 
"POSIXt"), tzone = "GMT"), What = structure(c(1L, 1L, 1L, 1L, 
1L, 1L), .Label = c("Dive", "Message", "Surface"), class = "factor"), 
    Shape = structure(c(2L, 4L, 3L, 2L, 2L, 2L), .Label = c("", 
    "Square", "U", "V"), class = "factor"), DepthMean = c(14.5, 
    16.5, 13, 14.5, 11, 12.5), DurationMean = c(76, 54, 40, 126, 
    198, 62), DepthMin = c(14.5, 16.5, 13, 14.5, 11, 12.5), DepthMax = c(14.5, 
    16.5, 13, 14.5, 11, 12.5), depth_range = structure(c(1L, 
    1L, 1L, 1L, 1L, 1L), .Label = c("shallow", "deep"), class = c("ordered", 
    "factor")), MidTime = structure(c(1477323906, 1477323973, 
    1477324022, 1477324107, 1477324359, 1477324511), class = c("POSIXct", 
    "POSIXt"), tzone = "GMT"), year = c(2016, 2016, 2016, 2016, 
    2016, 2016), id = c("111868_16", "111868_16", "111868_16", 
    "111868_16", "111868_16", "111868_16"), segmentid = c("111868_16", 
    "111868_16", "111868_16", "111868_16", "111868_16", "111868_16"
    ), mu.x = c(-4446545.25191192, -4446557.10576816, -4446565.77504969, 
    -4446580.81370994, -4446625.40007808, -4446652.29459533), 
    mu.y = c(-2305423.86124176, -2305461.88537725, -2305489.69364377, 
    -2305537.93137917, -2305680.93056743, -2305767.17264774), 
    lon = c(-39.9439956132156, -39.944102098218, -39.944179975699, 
    -39.9443150702825, -39.9447155964422, -39.9449571940013), 
    lat = c(-20.3985940756941, -20.3989161274532, -20.3991516537744, 
    -20.3995602097098, -20.4007713539709, -20.4015017842338), 
    lq_closest_filt = c(7L, 7L, 7L, 7L, 7L, 7L), dt_closest_filt = c(0.0516666666666667, 
    0.0702777777777778, 0.0838888888888889, 0.1075, 0.1775, 0.219722222222222
    ), dist_closest_filt = c(0.103680210832692, 0.141026573116106, 
    0.168339162761167, 0.215717097671267, 0.356168027785347, 
    0.440874049523752), rel.angle = c(NA_real_, NA_real_, NA_real_, 
    NA_real_, NA_real_, NA_real_), speed = c(NA_real_, NA_real_, 
    NA_real_, NA_real_, NA_real_, NA_real_), depth_bin = structure(c(1L, 
    1L, 1L, 1L, 1L, 1L), .Label = c("(0,50]", "(50,100]", "(100,150]", 
    "(150,200]", "(200,250]", "(250,300]", "(300,350]", "(350,400]", 
    "(400,450]", "(450,500]", "(500,550]", "(550,600]", "(600,650]", 
    "(650,700]"), class = "factor"), bat = structure(list(depth = c(-59L, 
    -59L, -59L, -59L, -59L, -59L)), row.names = c(NA, 6L), class = "data.frame"), 
    area = c("AR", "AR", "AR", "AR", "AR", "AR")), row.names = c(NA, 
6L), class = "data.frame") 

有人知道该如何解决？谢谢！

Answer 1

听起来您可能希望采用一些规则来决定哪些AM行变为AR。

• 如果连续的AM的数量是<20
• 如果以下目的地是不是 AA

一种方法是使用rle添加与这两个规则相关的列。对于重复序列中的连续值数，一列将具有lengths。另一列将具有“下一个”区域。这对于确定目的地是返回繁殖区还是返回饲养区至关重要。

最后，您可以使用条件语句，并将满足以下条件的行从AM更改为AR：

• 当前area为AM
• 之后的下一个area是不是 AA
• 重复值的数量小于20

这里是代码：

df2 <- cbind(df, next_area = with(rle(df$area), rep(c(values[-1], NA), lengths)), count = with(rle(df$area), rep(lengths, lengths)))
df2$area <- ifelse(with(df2, area == "AM" & next_area != "AA" & count < 20), "AR", df2$area)