【问题标题】:Calculate mean for subset of column such as applicationdate < max(applicationdate) using Data.Table in R使用 R 中的 Data.Table 计算列子集的平均值,例如 applicationdate < max(applicationdate)
【发布时间】:2021-08-21 15:40:48
【问题描述】:

我正在尝试创建一个新列,我想在其中计算历史数据的平均值,然后将其与我最近的数据点进行比较。

这是我的桌子的快照

假设我有第 15 周到第 22 周的数据点。我只想计算第 15 周到第 21 周的缺失数据点百分比的平均值。现在我必须在第 15-21 周对表进行切片,计算平均值,然后将值连接回来。我可以在 1 行代码中完成所有这些操作吗?

这是输入 结构(列表(ApplicationDate_Week = c(“2021-22”,“2021-21”, “2021-20”、“2021-19”、“2021-18”、“2021-17”、“2021-16”、“2021-15”、 "2021-14", "2021-13"), GNSVIPPWorkPhone_CountTidy = c(-0.233357452966715, -0.209491850431448,-0.162445638270657,-0.144578313253012,-0.127196433254655, -0.172413793103448,-0.0872703412073491,-0.0676691729323308, -0.0922794117647059, -0.045933734939759), GNSVIPPVIPLevelTidy = c(11.6577424023155, 11.3624161073826, 11.4561268866718, 11.1899691790417, 11.5022292158405, 11.5555204049351, 11.3041338582677, 10.8832796276405, 11.1113970588235, 10.7146084337349), GNSVIPPScore = c(0.759135636758321, 0.752135714285714, 0.754619109746738, 0.757030680862987, 0.76484948859166, 0.76503973426131, 0.766873261154856, 0.767475008950949, 0.773912169117647, 0.775240888554217 ), GNSVIPPREQUESTEMOUNT = c(1116.56150506512, 1126.86960690316, 1113.65464313124、1135.49313533203、1083.04720692368、1124.04334071496、 1130.33956692913、1135.08664518439、1078.76838235294、1090.27635542169 ), GNSVIPPRatioPaidLeadShoppers6M = c(42.2476360874848, 40.5152291014015, 43.8557045021645, 41.7211306948919, 39.5309453395192, 36.7297035590707, 36.0122896923077, 35.9960371273713, 34.5639904451683, 32.1152020044543 ), GNSVIPPRatioPaidLeadShoppers3Y = c(6.09824186145346, 6.04561388420461, 6.8275217979555, 6.73283313706248, 6.53066157323689, 6.06858887288441, 6.45259969890854, 6.75444378151261, 6.65001064182195, 6.42099029850746 ), GNSVIPPRatioPaidLeadApplied6M = c(0.4577996600085, 0.453706801573918, 0.467274894768491, 0.476851455675499, 0.456196172393008, 0.46552844796543, 0.471550018818216, 0.506900084033613, 0.492096231884058, 0.501193810359965 ), GNSVIPPRatioPaidLeadApplied3Y = c(0.691311602209945, 0.709030354131535, 0.704761455201443, 0.709077886817141, 0.681294424352019, 0.683463917897011, 0.692050056454648, 0.698919369747899, 0.691015403726708, 0.686982396839333 ), GNSVIPPeriodicity_CountTidy = c(-0.664254703328509, -0.64046021093001, -0.625735482220517,-0.591762398430933,-0.639391555205875,-0.61942423283771, -0.596456692913386,-0.544575725026853,-0.550735294117647,-0.565512048192771 ), GNSVIPPMonthDayOfApp = c(9.16823444283647, 26.7440076701822, 18.9634177539013, 12.0154104791258, 4.88696564384999, 25.3954444795951, 20.9396325459318, 14.0171858216971, 6.97426470588235, 17.3542921686747 ), GNSVIPPNumTimesApplied3Y = c(13.5873353166171, 13.1888701517707, 14.129885748647, 14.1586522734707, 14.6503918022905, 14.2563917897011, 15.2299585999247、15.5415966386555、15.2596273291925、15.8007023705004 ), GNSVIPPNumTimesFunded3Y = c(4.2422439439014, 4.00730747611017, 4.06043295249549, 3.984952567877, 4.41500904159132, 4.3035649981995, 4.4264207753105、4.45420168067227、4.47163561076605、4.71949078138718 ), GNSVIPPNumTimesFundedUnpaid6M = c(0.814279643008925, 0.745362563237774, 0.726097414311485, 0.757278377494275, 0.871609403254973, 0.868563197695355, 0.9028980052691, 0.916806722689076, 0.923809523809524, 1.0531167690957 ), GNSVIPPMarketingCategoryLeadGen = c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0), GNSVIPPMilitaryBank = c(0.0112156295224313, 0.011025886864813, 0.0176515732924021, 0.011207621182404, 0.013375295043273, 0.012337867763366, 0.0144356955380577, 0.0139634801288937, 0.0143382352941176, 0.0120481927710843 ), GNSVIPPage = c(48.331982633864, 47.8451581975072, 48.4526221540036, 48.3118520594004, 48.9297665879885, 48.0907624169567, 48.6264107611549, 49.1156820622986, 49.0780882352941, 49.1096762048193), GNSVIPPAgeEmpl = c(22.9432344428365, 21.9675934803452、21.9929649526733、22.5007565144298、22.6497246262785、 22.9768744068333, 22.7043635170604, 22.1306838524884, 22.8651102941176, 21.7357680722892), GNSVIPPApplNo_CountTidy = c(2.39182344428365, 2.38830297219559, 2.76745970836531, 2.77416643317456, 2.86388670338316, 2.84024043024359, 3.18208661417323, 3.31220909416398, 3.23970588235294, 3.42018072289157), GNSVIPPAppStore = c(15.9855282199711, 14.7430488974113, 15.3220772576106、14.9918744746428、16.4379753474954、16.6662448592218、 16.5682414698163、16.2169709989259、16.6871323529412、16.8930722891566 ), GNSVIPPCreditUnion = c(0.0734442836468886, 0.0781399808245446, 0.0874904067536454, 0.0795741103950686, 0.0823498557566221, 0.0816197405884214, 0.077755905511811, 0.0930898675259578, 0.0823529411764706, 0.0775602409638554 ), GNSVIPPCUSTEMAIL_FName_LV = c(0.296136685962373, 0.304079002876318, 0.293889741621898, 0.29356758195573, 0.292182323629688, 0.295965042708004, 0.29993562992126, 0.301187218045113, 0.291010992647059, 0.298249774096386 ), GNSVIPPCUSTEMAIL_LName_LV = c(0.358123191027496, 0.355575503355705, 0.357987081094909, 0.351115802745867, 0.354893233674272, 0.359659727934198, 0.36362125984252, 0.368595345506624, 0.364705551470588, 0.362464595103578 ), GNSVIPPDateFromDateRequestedEffectiveDate = c(1.56657018813314, 2.68648130393097、1.75748273215656、1.76576071728776、1.73118279569892、 1.81999367288833, 2.09744094488189, 1.78338703902614, 1.75661764705882, 1.74924698795181), GNSVIPPDaysFromLastPayDay = c(5.3321554770318, 5.95855224751897、6.79184066816576、6.24417391304348、5.42245989304813、 7.01114827201784, 7.34013605442177, 7.12942191544435, 6.48062360801782, 7.36672710788758), GNSVIPPDaysSinceLastWithdrawTidy = c(29.5900868306802, 26.934803451582, 33.6792018419033, 30.5673858223592, 29.7474429583005, 29.3669724770642, 31.4311023622047, 27.3766559255281, 30.5569852941176, 32.0395331325301), GNSVIPPDaysTillPayDay = c(11.3263386396527, 10.3005752636625, 10.1621898183679, 11.1045110675259, 12.0920535011802, 10.5855741853844, 9.99179790026247, 10.8496240601504, 11.7621323529412, 11.5670180722892), GNSVIPPFirstPmtDefaultCountTidy = c(-0.935238784370478, -0.898849472674976,-0.926579687899719,-0.91622303166153,-0.920272751114608, -0.937361594432142,-0.918963254593176,-0.915861081274615,-0.919485294117647, -0.949171686746988), GNSVIPPFundRate3Y = c(0.463055631109222, 0.453344406970208, 0.440381629585087, 0.448397317631665, 0.45974415310428, 0.46495855239467, 0.446069100489274, 0.449124159663866, 0.45170364389234, 0.44867181738367), GNSVIPPIncome_CountTidy = c(0.175470332850941, 0.226270373921381, 0.298030186748529, 0.307369010927431, 0.292682926829268, 0.333755140778235, 0.371719160104987, 0.421768707482993, 0.415073529411765, 0.490963855421687), GNSVIPPIncomeCapped = c(4248.42668567294, 4107.16932253116, 4279.30264548478, 4220.00788517792, 4185.8086748492, 4171.01028380259、4206.05106975066、4219.50144343、4126.59607463235、 4198.93875990211), GNSVIPPLendProtectScore = c(695.453854505972, 694.060402684564, 695.019698132515, 695.013733183857, 695.360535292574, 694.114873417721, 695.047244094488, 695.157894736842, 694.897426470588, 694.640060240964), GNSVIPPLendProtectScorePrior = c(699.770547945206, 698.035512965051, 698.495778045838, 698.518858642178, 698.214328899637, 697.417118093174, 698.01055408971, 698.656276326874, 698.195760598504, 698.092192324658), GNSVIPPLoanToIncomeCapped = c(0.295205390738061, 0.318314765100671, 0.297501432591456, 0.307545278789577, 0.293941987936008, 0.308343530528314, 0.306673064304462, 0.310265127103473, 0.295478308823529, 0.296143335843373), GNSVIPPLPScoreDiff = c(-4.43811563169165, -4.26550169109357,-3.77171290711701,-3.76345144356955,-3.09467634603751, -3.64559248554913,-3.11345646437995,-3.71356360572873,-3.45137157107232, -3.68901632112925), RMLendProtectScorePrior = c(699.770547945206, 698.035512965051, 698.510548523207, 698.518858642178, 698.214328899637, 697.417118093174, 698.01055408971, 698.656276326874, 698.195760598504, 698.092192324658), RMNumTimesFunded6M = c(1.0012749681258, 0.960652051714446, 0.908353365384615, 0.968269545305855, 1.08077154912598, 1.09110550954267, 1.13586751975913、1.16218487394958、1.18633540372671、1.3305531167691 ), RMNumTimesApplied3Y = c(13.5873353166171, 13.1888701517707, 14.1277043269231, 14.1586522734707, 14.6503918022905, 14.2563917897011, 15.2299585999247、15.5415966386555、15.2596273291925、15.8007023705004 ), RMNumTimesApplied6M = c(3.19634509137272, 3.2079820123665, 3.43449519230769, 3.57212953876349, 3.59162145871007, 3.59776737486496, 3.91908167105758、4.15672268907563、3.99130434782609、4.22080772607551 ), RMNumTimesAppliedPaid3Y = c(9.80195495112622, 9.6424957841484, 10.4693509615385, 10.6208701341184, 10.6515973477999, 10.3510983075261, 11.1633421151675、11.4915966386555、11.1540372670807、11.4266900790167 ), RMNumTimesAppliedPaid6M = c(2.2163195920102, 2.26869027543564, 2.51983173076923、2.63002944062807、2.52200120554551、2.53871083903493、 2.78471960858111、3.03571428571429、2.80331262939959、2.95917471466198 ), RMNumTimesFunded3Y = c(4.2422439439014, 4.00730747611017, 4.06009615384615, 3.984952567877, 4.41500904159132, 4.3035649981995, 4.4264207753105、4.45420168067227、4.47163561076605、4.71949078138718 ), RMNumTimesFundedUnpaid6M = c(0.814279643008925, 0.745362563237774, 0.726862980769231, 0.757278377494275, 0.871609403254973, 0.868563197695355, 0.9028980052691, 0.916806722689076, 0.923809523809524, 1.0531167690957 )), row.names = c(NA, -10L), class= c("data.table", "data.frame" ), .internal.selfref = )

这是我当前的代码

#GNS mean all variable

    GNS_mean = GNS_var[, lapply(.SD, mean,na.rm = TRUE), by = ApplicationDate_Week][order(-ApplicationDate_Week)]
    
    current_week = current_week_missing

# Historical mean and sd of MEAN 

    GNS_mean_historical = GNS_mean[ApplicationDate_Week < current_week,]
    GNS_mean_hist_mean = GNS_mean_historical[,lapply(.SD,mean,na.rm=TRUE)]
    GNS_mean_hist_mean[is.na(GNS_mean_hist_mean )] <- "Historical mean"
    GNS_mean_hist_sd = GNS_mean_historical[,lapply(.SD,sd,na.rm=TRUE)]
    GNS_mean_hist_sd[is.na(GNS_mean_hist_sd )] <- "Historical sd"

#GNS mean most current week

    GNS_mean_current = GNS_mean[ApplicationDate_Week == current_week,]
    GNS_mean_flag = rbind(GNS_mean_current,GNS_mean_hist_mean,GNS_mean_hist_sd)

这是我当前的代码。平均值不是cum mean,它是每个变量的所有历史数据点的平均值

如果您还有其他需要,请告诉我

【问题讨论】:

  • 请使用dput函数提供您的数据,即dput(df)。
  • 您好,我是第一次使用,如何使用 dput 函数将数据从我的 R 提供给 stackoverflow?
  • 在控制台中运行dput(head(mydata,20))。复制以structure( 开头的输出,并确保您不跳过任何内容,我们需要整个内容才能使用它(这很有帮助)。然后edit您的问题并粘贴到code block。
  • 您好 Peace,我用 dput 和我的代码编辑了帖子。让我知道这是否有效

标签: r data.table mean


【解决方案1】:

如果我正确理解了这个问题,这可以满足您的需求:

library(data.table)
data <- data.table(dat = as.Date("2021-01-01")+1:10, value = 1:10)
data[ ,meanval:=mean(fifelse(dat==max(dat),NA_real_,value),na.rm=T)]
data
#>            dat value meanval
#>  1: 2021-01-02     1       5
#>  2: 2021-01-03     2       5
#>  3: 2021-01-04     3       5
#>  4: 2021-01-05     4       5
#>  5: 2021-01-06     5       5
#>  6: 2021-01-07     6       5
#>  7: 2021-01-08     7       5
#>  8: 2021-01-09     8       5
#>  9: 2021-01-10     9       5
#> 10: 2021-01-11    10       5
mean(1:9)
#> [1] 5

【讨论】:

    【解决方案2】:

    您可以使用cummean(累积平均值)和shift 的组合:

    library(data.table)
    DT <- data.table(W = c("2016-21","2016-22","2016-23","2016-24"), n_miss = c(0, 1, 1, 0))
    

    如果您希望当前行包含在历史平均值中,那么

    DT[, hist_mean := cummean(n_miss)]
    #          W n_miss hist_mean
    #     <char>  <num>     <num>
    # 1: 2016-21      0 0.0000000  # mean(c(0))
    # 2: 2016-22      1 0.5000000  # mean(c(0,1))
    # 3: 2016-23      1 0.6666667  # mean(c(0,1,1))
    # 4: 2016-24      0 0.5000000  # mean(c(0,1,1,0))
    

    如果你不想包含当前的观察,那么

    DT[, hist_mean := shift(cummean(n_miss))]
    #          W n_miss hist_mean
    #     <char>  <num>     <num>
    # 1: 2016-21      0        NA  # no history
    # 2: 2016-22      1 0.0000000  # mean(c(0))
    # 3: 2016-23      1 0.5000000  # mean(c(0,1))
    # 4: 2016-24      0 0.6666667  # mean(c(0,1,1))
    

    (cummean 的使用假定顺序很重要并且由外部控制。)

    【讨论】:

      猜你喜欢
      • 2021-12-05
      • 2012-03-09
      • 2014-08-09
      • 2019-01-27
      • 1970-01-01
      • 1970-01-01
      • 2017-10-04
      • 1970-01-01
      相关资源
      最近更新 更多