使用R查找两次访问之间的差异

Question

我有这样的数据框

vehicleId visitDate taskName
123       1/1/2013  Change Battery
456       1/1/2013  Wiper Blades Changed
123       1/2/2013  Tire Pressure Check
123       1/3/2013  Tire Rotation
456       3/1/2013  Tire Pressure Check

我想做的是

vehicleId visitDate timeBetweenVisits(hrs)
123        1/1/2013                     24
123        1/2/2013                    672
456        1/1/2013                     48

有什么想法可以使用R做到这一点吗？

Answer 1

加载和转换数据：

## data now comma-separated as you have fields containing whitespace
R> res <- read.csv(text="
vehicleId, visitDate, taskName
123,       1/1/2013,  Change Battery
456,       1/1/2013,  Wiper Blades Changed
123,       1/2/2013,  Tire Pressure Check
123,       1/3/2013,  Tire Rotation
456,       3/1/2013,  Tire Pressure Check", stringsAsFactors=FALSE)
R> res$visitDate <- as.Date(res$visitDate, "%m/%d/%Y")     ## now in Daye format

看它：

R> res
  vehicleId  visitDate               taskName
1       123 2013-01-01         Change Battery
2       456 2013-01-01   Wiper Blades Changed
3       123 2013-01-02    Tire Pressure Check
4       123 2013-01-03          Tire Rotation
5       456 2013-03-01    Tire Pressure Check
R> 

日期计算：

R> res[3,"visitDate"] - res[1,"visitDate"]
Time difference of 1 days
R> as.numeric(res[3,"visitDate"] - res[1,"visitDate"])
[1] 1
R> difftime(res[3,"visitDate"],res[1,"visitDate"], unit="hours")
Time difference of 24 hours
R> as.numeric(difftime(res[3,"visitDate"],res[1,"visitDate"], unit="hours"))
[1] 24
R> 

矢量化：

R> as.numeric(difftime(res[2:nrow(res),"visitDate"],
+                      res[1:(nrow(res)-1),"visitDate"], unit="hours"))
[1]    0   24   24 1368
R> 

您当然也可以将其分配给新列。 您可能还想通过车辆ID子集。

Answer 2

使用@Dirk的答案中的res ，这是一个可以完成工作的by表达式：

by(res, res$vehicleId, FUN=function(d)
                       {
                         data.frame(vehicleId=head(d$vehicleId, -1),
                                    visitDate=head(d$visitDate, -1), 
                                    tbv=diff(d$visitDate)) 
                       }
)
## res$vehicleId: 123
##   vehicleId  visitDate    tbv
## 1       123 2013-01-01 1 days
## 2       123 2013-01-02 1 days
## ---------------------------------------------------------------------------------------------- 
## res$vehicleId: 456
##   vehicleId  visitDate     tbv
## 1       456 2013-01-01 59 days