通过在 R 中没有唯一标识符的 id 和日期范围合并两个数据框来创建事件答案

【问题标题】：Create event by merging two data frames by id and date range with no unique identifier in R通过在 R 中没有唯一标识符的 id 和日期范围合并两个数据框来创建事件
【发布时间】：2020-05-12 14:08:28
【问题描述】：

对于临床研究分析，我想在 dataframe1 中创建一个事件，其中 dataframe2 的第一个日期位于 2 个日期之间，加/减 n（比如 7）天。问题是在数据框 1 中存在非唯一 ID。换句话说，如果 dataframe2（Datum）中有日期介于 Datum_Implant - 7days 和 Datum_Explant + 7 天之间，我想将这些案例合并到 dataframe1 （并在此创建一个“事件”）。我该怎么做？

这是我的数据框 1

> dput(y2)

结构（列表（PIZ = c（“15597283”，“15597283”，“15597283”，“15597283”， “20116404”、“25562151”、“25562151”、“25936299”、“27172679”），基准= c（“27.12.2004”，“27.12.2004”，“07.04.2005”，“12.05.2006”， NA, "29.12.2015", "22.01.2016", "13.12.2013", NA), Datum_Mibi = c(2004.98633879781, 2004.98633879781, 2005.26301369863, 2006.35890410959, NA, 2015.99178082192, 2016.05737704918, 2013.94794520548, NA), Keim = c("链球菌缓解", "链球菌缓解", "Koagulase neg.Staphylokokken", “Koagulase neg.Staphylokokken”，NA，“革兰氏阳性 Haufenkokken”， “表皮葡萄球菌（nach Anreicherung）”，“革兰氏阳性 Kettenkokken”， NA), 事件 = c(1, 1, 1, 1, NA, 1, 1, 1, NA), 改变 = c(NA, 14, 14, 15, 17, 10, 10, 10, 17), Geschlecht = c(NA, "M", “M”、“M”、“W”、“M”、“M”、“M”、“M”），Datum_Implant = c（NA， 2004.96721311475、2005.43561643836、2006.13698630137、2015.09863013699、 2015.90410958904、2016.0956284153、2013.28493150685、2015.29863013699 ), Datum_Explant = c(NA, 2005.26301369863, 2005.81095890411, NA, 2015.75068493151, 2016.05737704918, 2016.77322404372, 2014.31232876712, 2015.96164383562), KatheterNummer = c(NA, 2, 3, 4, 1, 1, 2, 1, 1)), class= c("data.table", "data.frame" ), row.names = c(NA, -9L), .internal.selfref = )

这个数据框 2

> dput(Labor_Neutrophile_alle2))



   structure(list(PIZ = c("20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "20116404", 
"20116404", "20116404", "20116404", "20116404", "20116404", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "25936299", "25936299", "25936299", "25936299", "25936299", 
"25936299", "27172679", "27172679", "27172679", "27172679", "27172679", 
"27172679", "27172679", "27172679", "27172679", "27172679", "27172679", 
"27172679", "27172679", "27172679", "27172679", "27172679", "27172679", 
"27172679", "27172679", "27172679", "27172679", "27172679", "27172679", 
"27172679", "27172679", "27172679", "27172679", "27172679", "27172679", 
"27172679", "27172679", "27172679", "27172679", "27172679", "27172679", 
"27172679", "27172679", "27172679", "27172679", "27172679", "27172679", 
"27172679", "27172679", "27172679", "27172679", "27172679", "27172679", 
"27172679", "27172679", "27172679", "27172679", "27172679", "27172679", 
"27172679", "27172679", "27172679", "27172679", "27172679", "27172679", 
"27172679", "27172679"), Datum = c(2015.09863013699, 2015.10684931507, 
2015.1095890411, 2015.17534246575, 2015.18630136986, 2015.19178082192, 
2015.19452054795, 2015.20821917808, 2015.21095890411, 2015.21369863014, 
2015.22191780822, 2015.23287671233, 2015.24383561644, 2015.24931506849, 
2015.25479452055, 2015.26301369863, 2015.27945205479, 2015.28493150685, 
2015.29863013699, 2015.32054794521, 2015.33698630137, 2015.36164383562, 
2015.37534246575, 2015.41369863014, 2015.44109589041, 2015.45205479452, 
2015.47945205479, 2015.50136986301, 2015.52054794521, 2015.5397260274, 
2015.5397260274, 2015.55068493151, 2015.55890410959, 2015.57260273973, 
2015.58630136986, 2015.61643835616, 2015.63835616438, 2015.66301369863, 
2015.66575342466, 2015.67397260274, 2015.68493150685, 2015.69315068493, 
2015.70684931507, 2015.74794520548, 2015.78904109589, 2015.86301369863, 
2016.00819672131, 2016.03825136612, 2016.11475409836, 2016.22404371585, 
2016.30601092896, 2016.36338797814, 2016.3825136612, 2016.41530054645, 
2016.60655737705, 2016.6693989071, 2016.72677595628, 2016.80327868852, 
2016.84153005464, 2016.86338797814, 2016.86885245902, 2016.87158469945, 
2016.87431693989, 2016.87978142077, 2016.88797814208, 2016.89617486339, 
2016.90710382514, 2016.91256830601, 2016.93169398907, 2016.93169398907, 
2016.93169398907, 2016.93715846995, 2016.94808743169, 2016.98633879781, 
2017.02739726027, 2017.06575342466, 2017.12054794521, 2017.23561643836, 
2017.50410958904, 2017.75890410959, 2017.96438356164, 2013.75890410959, 
2013.77808219178, 2013.79452054795, 2013.80273972603, 2013.81095890411, 
2013.82191780822, 2013.82739726027, 2013.8301369863, 2013.83561643836, 
2013.84109589041, 2013.84657534247, 2013.84931506849, 2013.90684931507, 
2013.92602739726, 2013.99452054795, 2014.00273972603, 2014.00821917808, 
2014.05205479452, 2014.07123287671, 2014.07945205479, 2014.09863013699, 
2014.11780821918, 2014.13698630137, 2014.15616438356, 2014.16712328767, 
2014.18630136986, 2014.2301369863, 2014.25205479452, 2014.26575342466, 
2014.28493150685, 2014.30410958904, 2014.32328767123, 2014.34246575342, 
2014.36438356164, 2014.4, 2014.42739726027, 2014.47397260274, 
2014.49589041096, 2014.61917808219, 2014.71506849315, 2014.73424657534, 
2014.78082191781, 2014.81095890411, 2014.84109589041, 2014.91506849315, 
2014.95616438356, 2014.99452054795, 2015.04109589041, 2015.0602739726, 
2015.06849315068, 2015.10684931507, 2015.14520547945, 2015.16438356164, 
2015.18356164384, 2015.22191780822, 2015.26301369863, 2015.31780821918, 
2015.43835616438, 2015.55890410959, 2015.64657534247, 2015.87671232877, 
2016.09016393443, 2016.35245901639, 2017.0904109589, 2018.16438356164, 
2018.4904109589, 2019.1397260274, 2019.44931506849, 2015.28493150685, 
2015.29863013699, 2015.30684931507, 2015.32328767123, 2015.34794520548, 
2015.35890410959, 2015.37534246575, 2015.38356164384, 2015.39726027397, 
2015.40821917808, 2015.43561643836, 2015.44109589041, 2015.45205479452, 
2015.4602739726, 2015.4602739726, 2015.51232876712, 2015.51232876712, 
2015.52876712329, 2015.56712328767, 2015.60547945205, 2015.64383561644, 
2015.70410958904, 2015.74246575342, 2015.76164383562, 2015.78082191781, 
2015.78630136986, 2015.8, 2015.81095890411, 2015.81917808219, 
2015.83835616438, 2015.84383561644, 2015.88219178082, 2015.91232876712, 
2015.92328767123, 2015.9397260274, 2015.99452054795, 2016.08743169399, 
2016.12568306011, 2016.24863387978, 2016.32513661202, 2016.41256830601, 
2016.4781420765, 2016.55191256831, 2016.62021857923, 2016.70491803279, 
2016.78142076503, 2016.87704918033, 2016.95628415301, 2017.03287671233, 
2017.1095890411, 2017.18630136986, 2017.27945205479, 2017.43561643836, 
2017.43561643836, 2017.47123287671, 2017.70410958904, 2017.92328767123, 
2018.04931506849, 2018.16438356164, 2018.29863013699, 2019.58356164384
), Wert = c(150, 110, 120, 2190, 5600, 3600, 4100, 1700, 1550, 
750, 750, 250, 400, 100, 250, 500, 750, 300, 600, 2400, 2500, 
1050, 2200, 1900, 1750, 3050, 2300, 2150, 9400, 5800, 6600, 1210, 
430, 40, 250, 200, 2500, 800, 1200, 1550, 300, 50, 1300, 1900, 
4800, 1450, 1750, 2000, 3050, 2050, 2500, 4600, 2600, 2500, 2050, 
1900, 1650, 3100, 1900, 1550, 1250, 820, 600, 930, 480, 320, 
380, 500, 570, 650, 550, 800, 1400, 2900, 2100, 1950, 1950, 2900, 
3000, 3050, 4800, 4300, 600, 2500, 300, 250, 20, 100, 10, 30, 
100, 440, 500, 1950, 1350, 300, 250, 800, 2400, 600, 200, 500, 
1150, 1100, 850, 1550, 8160, 200, 800, 250, 450, 1050, 2600, 
2500, 2100, 3400, 1850, 1750, 3100, 1700, 2200, 2000, 1050, 1300, 
2600, 2200, 1850, 1900, 1400, 1500, 1500, 900, 5500, 2200, 3400, 
1650, 2200, 2050, 2010, 2800, 2500, 3700, 2900, 2500, 3700, 5100, 
6300, 3050, 4050, 750, 560, 290, 970, 200, 450, 1240, 1200, 1500, 
300, 200, 300, 300, 390, 450, 770, 800, 1850, 3050, 1000, 1050, 
1300, 9050, 1930, 660, 390, 150, 100, 550, 2900, 2300, 700, 200, 
550, 1100, 1950, 750, 1250, 1550, 1150, 1700, 600, 1350, 2200, 
1350, 2050, 1300, 1800, 1500, 1900, 1200, 1800, 2100, 1700, 2300, 
1900, 2400, 1740, 2500, 2300, 2600)), row.names = c(NA, -210L
), class = c("tbl_df", "tbl", "data.frame"))

所以我的想法是通过“KatheterNummer”创建一个唯一标识符，但我找不到任何合理的解决方案。有点惭愧，我建议你这样做：

names(Labor_Neutrophile_alle2)[names(Labor_Neutrophile_alle2) == "Pat-ID"] <- "PIZ"
Labor_Neutrophile_alle2$Datum <- as.Date(Labor_Neutrophile_alle2$Datum, "%d.%m.%Y")
Labor_Neutrophile_alle2$Datum <- decimal_date(Labor_Neutrophile_alle2$Datum)


#a_all_000$Neutrophil <- ifelse(a_all_000$PIZ == Labor_Neutrophile_alle2$PIZ & Labor_Neutrophile_alle2$Datum > a_all_000$Datum_Implant & Labor_Neutrophile_alle2$Datum < a_all_000$Datum_Explant & Labor_Neutrophile_alle2$Wert > 500,1,0)

library(dplyr)

Labor_Neutrophile_alle2 <- arrange(Labor_Neutrophile_alle2, PIZ, Datum)

a11 <- y2 %>% filter(KatheterNummer == "1") %>% select(PIZ, Datum_Mibi, Datum_Implant, Datum_Explant, Datum_Mibi, Keim, Event, Alter, Geschlecht, KatheterNummer)
labor <- Labor_Neutrophile_alle2 %>% group_by(PIZ, Datum) %>% filter(Wert > 500) %>% select(PIZ, Datum, Wert)

a111 <- full_join(a11, Labor_Neutrophile_alle2, by="PIZ") %>% 
filter(Datum - 0.01917808219 >= Datum_Implant & Datum + 0.01917808219 <= Datum_Explant)
a111$Datum_Implant <- a111$Datum_Implant + 0.01917808219
a111$Datum_Explant <- a111$Datum_Implant - 0.01917808219

a111 <- a111 %>% 
  group_by(PIZ) %>% 
  filter(Wert > 500, row_number()==1)

a12 <- y2 %>% filter(KatheterNummer == "2") %>% select(PIZ, Datum_Mibi, Datum_Implant, Datum_Explant,  Datum_Mibi, Keim, Event, Alter, Geschlecht, KatheterNummer)

a112 <- full_join(a12, Labor_Neutrophile_alle2, by="PIZ") %>% 
filter(Datum - 0.01917808219 >= Datum_Implant & Datum + 0.01917808219 <= Datum_Explant)

a112$Datum_Implant <- a112$Datum_Implant + 0.01917808219
a112$Datum_Explant <- a112$Datum_Implant - 0.01917808219

a112 <- a112 %>% 
  group_by(PIZ) %>% 
  filter(Wert > 500, row_number()==1)

a13 <- y2 %>% filter(KatheterNummer == "3") %>% select(PIZ, Datum_Mibi, Datum_Implant, Datum_Explant, Datum_Mibi, Keim, Event, Alter, Geschlecht, KatheterNummer)

a113 <- full_join(a13, Labor_Neutrophile_alle2, by="PIZ") %>% 
filter(Datum - 0.01917808219 >= Datum_Implant & Datum + 0.01917808219 <= Datum_Explant)

a113$Datum_Implant <- a113$Datum_Implant + 0.01917808219
a113$Datum_Explant <- a113$Datum_Implant - 0.01917808219

a113 <- a113 %>% 
  group_by(PIZ) %>% 
  filter(Wert > 500, row_number()==1)

a14 <- y2 %>% filter(KatheterNummer == "4") %>% select(PIZ, Datum_Mibi, Datum_Implant, Datum_Explant,  Datum_Mibi, Keim, Event, Alter, Geschlecht, KatheterNummer)

a114 <- full_join(a14, Labor_Neutrophile_alle2, by="PIZ") %>% 
filter(Datum - 0.01917808219 >= Datum_Implant & Datum + 0.01917808219 <= Datum_Explant)

a114$Datum_Implant <- a114$Datum_Implant + 0.01917808219
a114$Datum_Explant <- a114$Datum_Implant - 0.01917808219

a114 <- a114 %>% 
  group_by(PIZ) %>% 
  filter(Wert > 500, row_number()==1)

a15 <- y2 %>% filter(KatheterNummer == "5") %>% select(PIZ, Datum_Mibi, Datum_Implant,  Datum_Explant, Datum_Mibi, Keim, Event, Alter, Geschlecht, KatheterNummer)

a115 <- left_join(a15, Labor_Neutrophile_alle2, by="PIZ") %>% 
filter(Datum - 0.01917808219 >= Datum_Implant & Datum + 0.01917808219 <= Datum_Explant)

a115$Datum_Implant <- a115$Datum_Implant + 0.01917808219
a115$Datum_Explant <- a115$Datum_Implant - 0.01917808219

a115 <- a115 %>% 
  group_by(PIZ) %>% 
  filter(Wert > 500, row_number()==1)

# alle
a_all_000 <- rbind(a11, a12, a13, a14, a15)
# alle labor, erstes mal > 500 zwischen impl, expl
a_all_111 <- rbind(a111, a112, a113, a114, a115) # nur 50 sind neutrophil?

a_all_1111 <- a_all_111 %>% filter(KatheterNummer == 1) %>% select(PIZ, Datum, Neutro_Wert = Wert, Neutrophilie, KatheterNummer)
a_all_1112 <- a_all_111 %>% filter(KatheterNummer == 2) %>% select(PIZ, Datum, Neutro_Wert = Wert, Neutrophilie, KatheterNummer)
a_all_1113 <- a_all_111 %>% filter(KatheterNummer == 3) %>% select(PIZ, Datum, Neutro_Wert = Wert, Neutrophilie, KatheterNummer)
a_all_1114 <- a_all_111 %>% filter(KatheterNummer == 4) %>% select(PIZ, Datum, Neutro_Wert = Wert, Neutrophilie, KatheterNummer)
a_all_1115 <- a_all_111 %>% filter(KatheterNummer == 5) %>% select(PIZ, Datum, Neutro_Wert = Wert, Neutrophilie, KatheterNummer)

a_all_11111 <- rbind(a_all_1111, a_all_1112, a_all_1113, a_all_1114, a_all_1115) 
...
...

我想要的结果是这样的：

> dput(Neutro)
structure(list(...1 = c(1, 2, 3, 3), PIZ = c(11364500, 11364500, 
11364500, 11364500), Datum_Mibi = c(2001.132, 2002.132, 2003.142, 
2003.152), Datum_Implant = c(2001.025, 2002.025, 2003.055, 2003.055
), Datum_Explant = c(2001.518, 2002.618, 2003.518, 2003.518), 
    Keim = c("Koagulase neg.Staphylokokken", "Koagulase ", NA, 
    "Koagulase "), Event = c(1, 1, 0, 1), Alter = c(13, 13, 14, 
    14), Katheter = c(1, 2, 3, 3), Neutrophilie = c(1, 0, 1, 
    1), Wert = c(1200, 3050, 3000, 3000)), row.names = c(NA, 
-4L), class = c("tbl_df", "tbl", "data.frame"))

【问题讨论】：

嗨 Bengalepunktcom，您能提供一个您想要的输出示例吗？
当然，伊恩，我做到了。我第一次做临床研究...
您提供的Labor_Neutrophile_alle2 子集中的任何值都与y2 的样本年份都不相同。我当然可以为您回答这个问题，但要做到这一点，我需要提供数据以重现您的预期输出。
好的，Ian，我选择了 3 个肯定出现在两个数据框中的 ID。伊恩，谢谢你的耐心。如果您对我如何自己学习这些东西有任何其他提示，请善待并告诉我。是否有任何关于临床研究的书籍/ youtube 视频。还是您提供在线课程？我要一个！

标签： r

【解决方案1】：

您正在寻找进行非 equi 连接。不幸的是，这对 dplyr 来说是具有挑战性的，但对于 data.table 来说很容易：

library(data.table)
setDT(Labor_Neutrophile_alle2)
setDT(y2)

首先，在 Neutrophil 数据上设置一个开始和结束窗口：

Labor_Neutrophile_alle2[,`:=`(Datum.Window.Start = Datum - 0.01917808219,
                              Datum.Window.End = Datum + 0.01917808219)]
Labor_Neutrophile_alle2
          PIZ    Datum Wert Datum.Window.Start Datum.Window.End
  1: 20116404 2015.099  150           2015.079         2015.118
  2: 20116404 2015.107  110           2015.088         2015.126
 ---                                                           
209: 27172679 2018.299 2300           2018.279         2018.318
210: 27172679 2019.584 2600           2019.564         2019.603

然后加入y2。

Labor_Neutrophile_alle2[y2,
                        on = .(PIZ,
                               Datum.Window.Start <= Datum_Implant,
                               Datum.Window.End >= Datum_Implant),
                        nomatch = 0,]
        PIZ    Datum Wert Datum.Window.Start Datum.Window.End i.Datum Datum_Mibi Keim Event Alter Geschlecht Datum_Explant KatheterNummer
1: 20116404 2015.099  150           2015.099         2015.099    <NA>         NA <NA>    NA    17          W      2015.751              1
2: 20116404 2015.107  110           2015.099         2015.099    <NA>         NA <NA>    NA    17          W      2015.751              1
3: 20116404 2015.110  120           2015.099         2015.099    <NA>         NA <NA>    NA    17          W      2015.751              1
4: 27172679 2015.285  750           2015.299         2015.299    <NA>         NA <NA>    NA    17          M      2015.962              1
5: 27172679 2015.299  560           2015.299         2015.299    <NA>         NA <NA>    NA    17          M      2015.962              1
6: 27172679 2015.307  290           2015.299         2015.299    <NA>         NA <NA>    NA    17          M      2015.962              1

您没有提供足够的样本数据，但希望这会让您朝着正确的方向前进。

【讨论】：

谢谢伊恩。这对我有用，也很有启发性。所以你用 7 行写了我可怜地试图用 100 行做的事情。没有提供更多数据的原因是由于帖子的字符限制，我根本不允许粘贴更多数据。我也无法“喜欢”您的答案，因为我还没有达到 15 分。但我保证，当我获得足够的积分时，我会这样做。
我很高兴它对你有用。不要对自己太苛刻。我已经在这方面工作了 10 多年，如果你坚持下去，我相信你也会变得更好。您可以通过单击答案左侧的灰色对勾来接受答案。