R 将特定列值拆分为多列
-数据集R 将特定列值拆分为多列,r,R,-数据集 ID<-c(1,2,3,4,5,6,7) method<-c("cheque","DD","DD","Cheque","NetBank","NetBank","Cash") type<-c("Type1","Type1","Type2","Type2","Type3","Type3","Type4") aid<-c("A1","A1","A2","A2","A3","A3","A4") month<-c("JAN","JAN","FEB","F
ID<-c(1,2,3,4,5,6,7)
method<-c("cheque","DD","DD","Cheque","NetBank","NetBank","Cash")
type<-c("Type1","Type1","Type2","Type2","Type3","Type3","Type4")
aid<-c("A1","A1","A2","A2","A3","A3","A4")
month<-c("JAN","JAN","FEB","FEB","MAR","MAR","APR")
year<-c(2016,2016,2015,2015,2017,2017,2018)
Outcome<-c("Positive","Positive","Negative","Negative","Medium","Medium","Neutral")
ser_no<-c("A00001","A00001","A00002","A00002","A00003","A00003","A00004")
Units<-c(100,200,300,400,500,600,700)
amt<-c(1000,1500,2000,3000,4000,2500,6000)
user_cnt<-c(20,20,15,15,32,32,44)
data<-data.frame(ID=ID,type=type,aid=aid,month=month,year=year,Outcome=Outcome,ser_no=ser_no,Units=Units,amt=amt,user_cnt=user_cnt,method=method)
- 我想在输出中添加来自
数据的方法列 方法列只能有四个值1.支票2.DD 3.NetBank 4.空白表示现金
1张支票
和1张DD
值,因此计数显示为1
<代码>净银行和现金
值不存在,因此计数为0。
根据GROUP BY子句,第三行有2个Netbank
值,因此计数显示为2
,并且没有Netbank
,Cash
和check
值,因此计数为0
type aid month year Outcome ser_no members entries UNITS amt LowestAmt HighestAmount Mean user_cnt Suggestion Cheque DD Netbank Cash
Type1 A1 JAN 2016 Positive A00001 2 2 300 2500 1000 1500 1250 20 0.10000000 1 1 0 0
Type2 A2 FEB 2015 Negative A00002 2 2 700 5000 2000 3000 2500 15 0.13333333 1 1 0 0
Type3 A3 MAR 2017 Medium A00003 2 2 1100 6500 2500 4000 3250 32 0.06250000 0 0 2 0
Type4 A4 APR 2018 Neutral A00004 1 1 700 6000 6000 6000 6000 44 0.02272727 0 0 0 1
使用数据表:
library(data.table)
DT <- setDT(data)
DT[,method := tolower(method)] # to avoid different count with upper and lower case
plouf<-dcast(DT[,.N, by = .(type,method)],type~ method)
plouf[is.na(plouf)]<-0
type cash cheque dd netbank
1: Type1 0 1 1 0
2: Type2 0 1 1 0
3: Type3 0 0 0 2
4: Type4 1 0 0 0
使用数据表:
library(data.table)
DT <- setDT(data)
DT[,method := tolower(method)] # to avoid different count with upper and lower case
plouf<-dcast(DT[,.N, by = .(type,method)],type~ method)
plouf[is.na(plouf)]<-0
type cash cheque dd netbank
1: Type1 0 1 1 0
2: Type2 0 1 1 0
3: Type3 0 0 0 2
4: Type4 1 0 0 0
我无法解决“支票”上的案例问题,因为tolower不在sqldf下工作。因此包括两个选项
sqldf("select type
,aid
,month
,year
,Outcome
,ser_no
,count(distinct ID) as members
,count(type) as entries
,sum(UNITS) as UNITS
,sum(amt) as amt
,min(amt) as LowestAmt
,max(amt) as HighestAmount
,AVG(amt) as Mean
,user_cnt
,cast (count(distinct ID) as real)/user_cnt as Suggestion
,count(case when lower(method)='cheque' then method end ) as cheque
,count(case when method ='DD' then method end ) as DD
,count(case when method ='NetBank' then method end ) as NetBank
,count(case when method ='Cash' then method end ) as Cash
from data
group by type,aid,month,year,Outcome,ser_no")
我无法解决“支票”上的案例问题,因为tolower不在sqldf下工作。因此包括两个选项
sqldf("select type
,aid
,month
,year
,Outcome
,ser_no
,count(distinct ID) as members
,count(type) as entries
,sum(UNITS) as UNITS
,sum(amt) as amt
,min(amt) as LowestAmt
,max(amt) as HighestAmount
,AVG(amt) as Mean
,user_cnt
,cast (count(distinct ID) as real)/user_cnt as Suggestion
,count(case when lower(method)='cheque' then method end ) as cheque
,count(case when method ='DD' then method end ) as DD
,count(case when method ='NetBank' then method end ) as NetBank
,count(case when method ='Cash' then method end ) as Cash
from data
group by type,aid,month,year,Outcome,ser_no")
整个聚合可以使用
数据在一条语句中完成。表:
library(data.table)
setDT(data)[
, .(members = uniqueN(ID), entries = .N, UNITS = sum(Units), amt = sum(amt),
LowestAmt = min(amt), HighestAmount = max(amt), Mean = mean(amt),
user_cnt = first(user_cnt), Suggestion = uniqueN(ID) / first(user_cnt),
Cheque = sum(tolower(method) == "cheque"), DD = sum(tolower(method) == "dd"),
NetBank = sum(tolower(method) == "netbank"),
Cash = sum(tolower(method) %in% c("cash", ""))),
by = .(type, aid, month, year, Outcome, ser_no)]
如果method
中有4个以上的不同值,我建议使用其他方法,如dcast()
和join 整个聚合可以使用数据在一条语句中完成。表
:
library(data.table)
setDT(data)[
, .(members = uniqueN(ID), entries = .N, UNITS = sum(Units), amt = sum(amt),
LowestAmt = min(amt), HighestAmount = max(amt), Mean = mean(amt),
user_cnt = first(user_cnt), Suggestion = uniqueN(ID) / first(user_cnt),
Cheque = sum(tolower(method) == "cheque"), DD = sum(tolower(method) == "dd"),
NetBank = sum(tolower(method) == "netbank"),
Cash = sum(tolower(method) %in% c("cash", ""))),
by = .(type, aid, month, year, Outcome, ser_no)]
如果method
中有4个以上的不同值,我建议使用其他方法,如dcast()
和join 有没有办法在一个查询本身中获得预期的输出,而不是附加它?我在六个列上分组,分别是类型、援助、月、年、结果、序号
,而不仅仅是类型
dcast(data,type+month+year+Outcome+aid+seru no~method,fun.aggregate=length)
您可以通过创建一个分组变量来使用相同的代码:DT[,grp:=粘贴(type,aid,month,year,Outcome,seru no,sep=“”)]
然后dcast(DT[,.N,by=(grp,method)],grp~method)
有没有办法在一个查询本身中获得预期的输出,而不是附加它?我在六个列上分组,分别是类型、援助、月、年、结果、序号
,而不仅仅是类型
dcast(data,type+month+year+Outcome+aid+seru no~method,fun.aggregate=length)
您可以通过创建一个分组变量来使用相同的代码:DT[,grp:=粘贴(type,aid,month,year,Outcome,seru no,sep=“”)]
然后dcast(DT[,.N,by=(grp,method)],grp~method)
在140万行的数据集上,它比PIG答案慢得多。清管器回答需要17秒,数据表需要1分14秒@在140万行的数据集上,UweIt比PIG答案慢得多。清管器回答需要17秒,数据表需要1分14秒@Uwe
type aid month year Outcome ser_no members entries UNITS amt LowestAmt HighestAmount Mean user_cnt Suggestion Cheque DD NetBank Cash
1: Type1 A1 JAN 2016 Positive A00001 2 2 300 2500 1000 1500 1250 20 0.10000000 1 1 0 0
2: Type2 A2 FEB 2015 Negative A00002 2 2 700 5000 2000 3000 2500 15 0.13333333 1 1 0 0
3: Type3 A3 MAR 2017 Medium A00003 2 2 1100 6500 2500 4000 3250 32 0.06250000 0 0 2 0
4: Type4 A4 APR 2018 Neutral A00004 1 1 700 6000 6000 6000 6000 44 0.02272727 0 0 0 1