我在下面提到了不同的数据帧:
DF1:
Origination_Date Count1 Count2
2018-07-01 147 205
2018-07-05 180 345
2018-07-08 195 247
2018-08-04 205 788
DF2:
Date ID
2018-07-01 I-1
2018-07-02 I-2
2018-07-02 I-3
2018-07-03 I-4
2018-07-03 I-5
2018-08-04 I-6
2018-08-04 I-7
支付
Create_Date ID
2018-07-01 I-1
2018-07-02 I-2
2018-07-03 I-3
2018-08-04 I-4
2018-08-04 I-5
通过利用上述多个数据帧,我想通过MonthYear创建一个新的数据帧组,并在月份和日期方面表示合并计数,如下面的示例数据框所示。
要求输出:
Month Count1 Count2 DF2_Count(ID) DF3_Count(ID)
Aug-18 205 788 2 2
Jul-18 522 797 5 3
Jun-18 0 0 0 0
上面提到的相同数据结构也希望在日期的基础上创建,我尝试使用group_by函数,并且可以为每个单独的datafreme创建所需的数据帧,但不能通过合并所有数据帧。
注意: - 虽然我的数据框中没有Jun-18
月,但我想在同一个月创建一行(想要在最近一个月创建至少三个月的所需输出数据帧(即如果它的Sep-18
比Aug-18
和Jul-18
) - 如果任何数据帧的0行比显示计数0,则为必需输出。
这样的事情怎么样:
# your data
df1 <- data.frame (Origination_Date = c('2018-07-01','2018-07-05','2018-07-08','2018-08-04'),
Count1 = c(147,180,195,205), Count2 = c(205,345,247,788))
df2 <- data.frame (Date = c('2018-07-01','2018-07-02','2018-07-02','2018-07-03','2018-07-03','2018-08-04','2018-08-04'),
ID = c('I-1','I-2','I-3','I-4','I-5','I-6','I-7'))
df3 <- data.frame (Create_Date = c('2018-07-01','2018-07-02','2018-07-03','2018-08-04','2018-08-04'), ID = c('I-1','I-2','I-3','I-4','I-5'))
# package to manage date
library(lubridate)
# first we create the yyyy-mm data.frame grouped
df1_1 <- df1 %>%
mutate(ym = format(ymd(Origination_Date),'%Y-%b')) %>%
group_by(ym) %>%
summarise(Count1 = sum(Count1) ,Count2 = sum(Count2))
df2_1 <- df2 %>%
mutate(ym = format(ymd(Date),'%Y-%b')) %>%
group_by(ym) %>%
summarise(DF2_Count = n())
df3_1 <- df3 %>%
mutate(ym = format(ymd(Create_Date),'%Y-%b')) %>%
group_by(ym) %>%
summarise(DF3_Count = n())
# join them together
df_1 <- df1_1 %>% full_join(df2_1, by = 'ym') %>% full_join(df3_1, by = 'ym')
> df_1
# A tibble: 2 x 5
ym Count1 Count2 DF2_Count DF3_Count
<chr> <dbl> <dbl> <int> <int>
1 2018-Aug 205 788 2 2
2 2018-Jul 522 797 5 3
现在是棘手的部分,添加缺少的月份,我创建了一对如果谁检查是否没有最大月 - 2 - (第二个),它添加一个假行,第一个为最后一个但是一个。
if(
format(floor_date(as.Date(max(union(union(df1[,1], df2[,1]),df3[,1]))), "month") - months(1),'%Y-%b') %in% df_1$ym == F){
df_2 <- data.frame(ym =format(floor_date(as.Date(max(union(union(df1[,1], df2[,1]),df3[,1]))), "month") - months(1),'%Y-%b'),
Count1 = 0,
Count2 = 0,
DF2_Count= 0,
DF3_Count= 0)
rbind(df_1,df_2)} else {'it already exists'}
[1] "it already exists"
if(
format(floor_date(as.Date(max(union(union(df1[,1], df2[,1]),df3[,1]))), "month") - months(2),'%Y-%b') %in% df_1$ym == F){
df_2 <- data.frame(ym =format(floor_date(as.Date(max(union(union(df1[,1], df2[,1]),df3[,1]))), "month") - months(2),'%Y-%b'),
Count1 = 0,
Count2 = 0,
DF2_Count= 0,
DF3_Count= 0)
rbind(df_1,df_2)
} else {'it already exists'}
# A tibble: 3 x 5
ym Count1 Count2 DF2_Count DF3_Count
<chr> <dbl> <dbl> <dbl> <dbl>
1 2018-Aug 205 788 2 2
2 2018-Jul 522 797 5 3
3 2018-Jun 0 0 0 0
这是data.table
的解决方案:
library(data.table)
DF1 <- fread(
"Origination_Date Count1 Count2
2018-07-01 147 205
2018-07-05 180 345
2018-07-08 195 247
2018-08-04 205 788")
DF2 <- fread(
"Date ID
2018-07-01 I-1
2018-07-02 I-2
2018-07-02 I-3
2018-07-03 I-4
2018-07-03 I-5
2018-08-04 I-6
2018-08-04 I-7")
DF3 <- fread(
"Create_Date ID
2018-07-01 I-1
2018-07-02 I-2
2018-07-03 I-3
2018-08-04 I-4
2018-08-04 I-5")
S1 <- DF1[, Ymon:=substr(Origination_Date, 1, 7)][, .(sum(Count1), sum(Count2)), Ymon]
S2 <- DF2[, Ymon:=substr(Date, 1, 7)][, .(DF2count=.N), Ymon]
S3 <- DF3[, Ymon:=substr(Create_Date, 1, 7)][, .(DF3count=.N), Ymon]
S <- merge(data.table(Ymon=paste0("2018-0", 6:8)), S1, all.x=TRUE)
S <- merge(S, S2, all.x=TRUE)
S <- merge(S, S3, all.x=TRUE)
S
# > S
# Ymon V1 V2 DF2count DF3count
# 1: 2018-06 NA NA NA NA
# 2: 2018-07 522 797 5 3
# 3: 2018-08 205 788 2 2
如果你想要0
而不是NA
,你可以这样做:
S[is.na(S)] <- 0
S
# Ymon V1 V2 DF2count DF3count
# 1: 2018-06 0 0 0 0
# 2: 2018-07 522 797 5 3
# 3: 2018-08 205 788 2 2