1

我有一个数据框，其中包含课程 ID、学生 ID、周数（第一周为 1，第二周为 2，...），以及每个用户每周在每门课程中所做的一些信息。如果讲师在那周在该课程中“干预”学生，则 df 的最后两列是非 NA，否则为 NA。我想比较每个学生在第一次干预之前和之后的行为。

所以我想做的是一个专栏，“HasIntervened”，它比学生的第一次干预少几周为假，大于或等于几周为真，但我有一个地狱般的时间来创建它简单的列。我相当肯定这将aggregate是要走的路，但我只是没有以正确的方式思考这个问题。

这是dput数据框的前 60 行（5 个学生的价值）：

structure(list(UserID = c(4188948L, 4188948L, 4188948L, 4188948L, 
4188948L, 4188948L, 4735684L, 4735684L, 4735684L, 4735684L, 4735684L, 
4735684L, 6292486L, 6292486L, 6292486L, 6292486L, 6292486L, 6292486L, 
6469671L, 6469671L, 6469671L, 6469671L, 6469671L, 6469671L, 6538263L, 
6538263L, 6538263L, 6538263L, 6538263L, 6538263L, 6621258L, 6621258L, 
6621258L, 6621258L, 6621258L, 6621258L, 6891869L, 6891869L, 6891869L, 
6891869L, 6891869L, 6891869L, 6891869L, 6891869L, 6891869L, 6891869L, 
6891869L, 6891869L, 6978155L, 6978155L, 6978155L, 6978155L, 6978155L, 
6978155L, 7195846L, 7195846L, 7195846L, 7195846L, 7195846L, 7195846L
), CourseID = c(6567871L, 6567871L, 6567871L, 6567871L, 6567871L, 
6567871L, 6567168L, 6567168L, 6567168L, 6567168L, 6567168L, 6567168L, 
6567864L, 6567864L, 6567864L, 6567864L, 6567864L, 6567864L, 6567159L, 
6567159L, 6567159L, 6567159L, 6567159L, 6567159L, 6567162L, 6567162L, 
6567162L, 6567162L, 6567162L, 6567162L, 6567853L, 6567853L, 6567853L, 
6567853L, 6567853L, 6567853L, 6567159L, 6567159L, 6567159L, 6567159L, 
6567159L, 6567159L, 6567864L, 6567864L, 6567864L, 6567864L, 6567864L, 
6567864L, 6567873L, 6567873L, 6567873L, 6567873L, 6567873L, 6567873L, 
6567859L, 6567859L, 6567859L, 6567859L, 6567859L, 6567859L), 
WeekInCourse = c(1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 
3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 
4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 
5, 6, 1, 2, 3, 4, 5, 6), WeekPostCount = c(1L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 5L, 3L, 4L, 3L, 3L, 0L, 4L, 
0L, 0L, 0L, 0L, 0L, 3L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 1L, 0L, 0L, 0L, 0L, 2L, 
2L, 0L, 0L, 4L, 0L, 3L, 0L, 3L, 0L, 0L, 0L), WeekLoginCount = c(2L, 
1L, 0L, 0L, 0L, 0L, 1L, 1L, 1L, 0L, 0L, 0L, 4L, 4L, 1L, 0L, 
0L, 0L, 3L, 3L, 1L, 0L, 0L, 0L, 2L, 1L, 0L, 0L, 0L, 0L, 1L, 
1L, 0L, 0L, 0L, 0L, 3L, 0L, 0L, 0L, 0L, 0L, 4L, 1L, 0L, 0L, 
0L, 0L, 3L, 3L, 1L, 0L, 0L, 0L, 1L, 1L, 0L, 0L, 0L, 0L), 
WeekPointsPercent = c(0, 0, 0, 0, 0, 0, 0, 0.185714285714286, 
0.375, 0.2, 0, 0, 0, 0.85, 0.7, 0.4, 0.7, 0.7, 0, 0.857142857142857, 
0.35, 0, 0, 0.712765957446808, 0, 1, 0, 0, 0, 0, 0, 0, 0, 
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0.25, 0, 0, 0, 0, 0, 0.5, 0.5, 
0, 0, 0.7, 1, 1, 0.375, 0.723076923076923, 0, 0.738636363636364
), CumulativePointsPercent = c(0, 0, 0, 0, 0, 0, 0, 0.185714285714286, 
0.254545454545455, 0.235294117647059, 0.235294117647059, 
0.10958904109589, 0, 0.85, 0.8, 0.533333333333333, 0.55, 
0.563636363636364, 0, 0.857142857142857, 0.623076923076923, 
0.476470588235294, 0.476470588235294, 0.600558659217877, 
0, 1, 0.0666666666666667, 0.0666666666666667, 0.0461538461538462, 
0.0461538461538462, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 
0.25, 0.166666666666667, 0.0555555555555556, 0.05, 0.0454545454545455, 
0, 0.5, 0.5, 0.166666666666667, 0.15, 0.2, 1, 1, 0.615384615384615, 
0.669230769230769, 0.621428571428571, 0.666666666666667), 
RiskEstimate = c(0.627717786405816, 0.986868933315635, 0.986687587608184, 
0.993909863003438, 0.997123961252086, 0.995862152216296, 
0.914011371723269, 0.925359536086114, 0.902625588346349, 
0.956922151061089, 0.977244888475535, 0.975006380719003, 
0.215420992232115, 0.174623555825523, 0.241380495376484, 
0.699712463799006, 0.692014530298594, 0.697966901130338, 
0.765071150059092, 0.763071307309743, 0.767261726128078, 
0.835918063362269, 0.854949153314029, 0.805318343915736, 
0.792873572656207, 0.790581615380765, 0.82622599277251, 0.9330287497742, 
0.965763061363497, 0.951226314109191, 0.851355921713566, 
0.991081300877175, 0.989671569185701, 0.995402298000919, 
0.997671718747865, 0.996593366142757, 0.738690043138604, 
0.865412845144037, 0.831369850200541, 0.93845410260835, 0.968400480533385, 
0.9533338828382, 0.624930735381371, 0.981915016747928, 0.985037736895337, 
0.994680902796769, 0.996907588471311, 0.995388109404559, 
0.887995464972052, 0.970620002831325, 0.97136665697772, 0.992618626388727, 
0.99543249839328, 0.992149889176406, 0.923802324633255, 0.984464950934932, 
0.978726967214146, 0.971473084822075, 0.97886220009245, 0.979311013989987
), RiskBin = c(3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 
3L, 1L, 1L, 1L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 
3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 
3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 
3L, 3L, 3L, 3L), InterventionID = c(NA, 26L, NA, NA, NA, 
NA, NA, NA, NA, NA, 50L, NA, NA, NA, NA, NA, 73L, NA, NA, 
NA, NA, NA, 56L, NA, NA, NA, NA, 46L, NA, NA, NA, 33L, NA, 
NA, NA, NA, 15L, NA, NA, 43L, 53L, NA, NA, NA, NA, NA, 71L, 
NA, NA, NA, NA, NA, 78L, NA, NA, 36L, NA, NA, 80L, NA), InterventionType = structure(c(NA, 
2L, NA, NA, NA, NA, NA, NA, NA, NA, 3L, NA, NA, NA, NA, NA, 
2L, NA, NA, NA, NA, NA, 3L, NA, NA, NA, NA, 3L, NA, NA, NA, 
2L, NA, NA, NA, NA, 3L, NA, NA, 3L, 2L, NA, NA, NA, NA, NA, 
2L, NA, NA, NA, NA, NA, 2L, NA, NA, 3L, NA, NA, 3L, NA), .Label = c("", 
"At-Risk Form", "Email", "Other", "Phone"), class = "factor")), .Names = c("UserID", 
"CourseID", "WeekInCourse", "WeekPostCount", "WeekLoginCount", 
"WeekPointsPercent", "CumulativePointsPercent", "RiskEstimate", 
"RiskBin", "InterventionID", "InterventionType"), row.names = c(NA, 
60L), class = "data.frame")
4

4 回答 4

2 
courses$HasIntervened <- as.logical( with(courses, ave(InterventionID,
                                                 UserID, CourseID,  # grouping factors
                                           FUN=function(x) cumsum( !is.na(x) ) ) ) )
于 2012-10-09T23:24:25.090 回答
1

试试这个：

foo = your.data
foo$WeekInCourse[is.na(foo$InterventionID)]=Inf
bar = setNames(aggregate(WeekInCourse ~ UserID, foo, min),c("UserID","FirstW"))
foo = merge(foo, bar, by="UserID")
your.data$HasIntervened = your.data$WeekInCourse >= foo$FirstW
于 2012-10-09T23:20:12.603 回答
1

一种data.table编码优雅和内存效率的方法

library(data.table)
# assuming your data is in DF
DT <- as.data.table(DF)
# set the key to ensure that the data is sorted by week within 
# each user / course combination
setkey(DT, UserID, CourseID,  WeekInCourse)
# using cumsum  
DT[,hasIntervened := cumsum(!is.na(InterventionID))>0 ,by =list(CourseID, UserID)]

语法避免了data.table需要with

于 2012-10-16T04:20:28.527 回答
0

这应该有效：

library(plyr)
ddply(df, .(UserID), function(x) {
     i <- which.min((x$InterventionID))
    if(i>1) {
        x$HasIntervened <- c(rep(FALSE,i-1), rep(TRUE, nrow(x)-i+1))
        } else {
        x$HasIntervened <- TRUE     
        }
     x
    })
于 2012-10-10T00:59:00.593 回答