Thank you. I was just curious what sort=FALSE had no impact. Wondering what it is there for then... On Wed, Mar 8, 2017 at 11:43 AM, Jeff Newmiller <jdnewmil at dcn.davis.ca.us> wrote:> Merging is not necessarily an order-preserving operation, but sorting can make the operation more efficient. The sort=TRUE argument forces the result to be sorted, but sort=FALSE is in not a promise that order will be preserved. (I think the imperfect sorting occurs when there are multiple keys but am not sure.) You can add columns to the input data that let you restore some semblance of the original ordering afterward, or you can roll your own possibly-less-efficient merge using match and indexing: > > info[ match( grades2$grade, info$grade ), ] > -- > Sent from my phone. Please excuse my brevity. > > On March 8, 2017 8:07:27 AM PST, Dimitri Liakhovitski <dimitri.liakhovitski at gmail.com> wrote: >>Hello! >>I have a vector 'grades' and a data frame 'info': >> >>grades2 <- data.frame(grade = c(1,2,2,3,1)) >>info <- data.frame( >> grade = 3:1, >> desc = c("Excellent", "Good", "Poor"), >> fail = c(F, F, T) >>) >> >>I want to get the info for all grades I have in info: >> >>This solution resorts everything in the order of column 'grade': >>merge(grades2, info, by = "grade", all.x = T, all.y = F) >> >>Could you please explain why this solution also resorts - despite sort >>= FALSE? >>merge(grades2, info, by = "grade", all.x = T, all.y = F, sort = FALSE) >> >>Thanks a lot!-- Dimitri Liakhovitski
If you are still wondering, try re-reading my answer. FALSE is more efficient, TRUE is sorted. Lack of sorting has nothing to do with preserving order. -- Sent from my phone. Please excuse my brevity. On March 8, 2017 8:55:06 AM PST, Dimitri Liakhovitski <dimitri.liakhovitski at gmail.com> wrote:>Thank you. I was just curious what sort=FALSE had no impact. >Wondering what it is there for then... > >On Wed, Mar 8, 2017 at 11:43 AM, Jeff Newmiller ><jdnewmil at dcn.davis.ca.us> wrote: >> Merging is not necessarily an order-preserving operation, but sorting >can make the operation more efficient. The sort=TRUE argument forces >the result to be sorted, but sort=FALSE is in not a promise that order >will be preserved. (I think the imperfect sorting occurs when there are >multiple keys but am not sure.) You can add columns to the input data >that let you restore some semblance of the original ordering afterward, >or you can roll your own possibly-less-efficient merge using match and >indexing: >> >> info[ match( grades2$grade, info$grade ), ] >> -- >> Sent from my phone. Please excuse my brevity. >> >> On March 8, 2017 8:07:27 AM PST, Dimitri Liakhovitski ><dimitri.liakhovitski at gmail.com> wrote: >>>Hello! >>>I have a vector 'grades' and a data frame 'info': >>> >>>grades2 <- data.frame(grade = c(1,2,2,3,1)) >>>info <- data.frame( >>> grade = 3:1, >>> desc = c("Excellent", "Good", "Poor"), >>> fail = c(F, F, T) >>>) >>> >>>I want to get the info for all grades I have in info: >>> >>>This solution resorts everything in the order of column 'grade': >>>merge(grades2, info, by = "grade", all.x = T, all.y = F) >>> >>>Could you please explain why this solution also resorts - despite >sort >>>= FALSE? >>>merge(grades2, info, by = "grade", all.x = T, all.y = F, sort >FALSE) >>> >>>Thanks a lot!
Hello, If you need to preserve the order you can do it like this. inx <- order(grades2$grade) result <- merge(grades2, info, by = "grade", all.x = T, all.y = F, sort = FALSE) result[order(inx), ] Hope this helps, Rui Barradas Em 08-03-2017 16:55, Dimitri Liakhovitski escreveu:> Thank you. I was just curious what sort=FALSE had no impact. > Wondering what it is there for then... > > On Wed, Mar 8, 2017 at 11:43 AM, Jeff Newmiller > <jdnewmil at dcn.davis.ca.us> wrote: >> Merging is not necessarily an order-preserving operation, but sorting can make the operation more efficient. The sort=TRUE argument forces the result to be sorted, but sort=FALSE is in not a promise that order will be preserved. (I think the imperfect sorting occurs when there are multiple keys but am not sure.) You can add columns to the input data that let you restore some semblance of the original ordering afterward, or you can roll your own possibly-less-efficient merge using match and indexing: >> >> info[ match( grades2$grade, info$grade ), ] >> -- >> Sent from my phone. Please excuse my brevity. >> >> On March 8, 2017 8:07:27 AM PST, Dimitri Liakhovitski <dimitri.liakhovitski at gmail.com> wrote: >>> Hello! >>> I have a vector 'grades' and a data frame 'info': >>> >>> grades2 <- data.frame(grade = c(1,2,2,3,1)) >>> info <- data.frame( >>> grade = 3:1, >>> desc = c("Excellent", "Good", "Poor"), >>> fail = c(F, F, T) >>> ) >>> >>> I want to get the info for all grades I have in info: >>> >>> This solution resorts everything in the order of column 'grade': >>> merge(grades2, info, by = "grade", all.x = T, all.y = F) >>> >>> Could you please explain why this solution also resorts - despite sort >>> = FALSE? >>> merge(grades2, info, by = "grade", all.x = T, all.y = F, sort = FALSE) >>> >>> Thanks a lot! > > >
I understood your answer. The point is that sort = TRUE that doesn't sort is plain confusing. Instead, the option should have been something like efficient = TRUE or FALSE. At least then no one would stupidly expect sort = TRUE to sort and sort = FALSE to NOT sort. On Wed, Mar 8, 2017 at 12:51 PM, Jeff Newmiller <jdnewmil at dcn.davis.ca.us> wrote:> If you are still wondering, try re-reading my answer. FALSE is more efficient, TRUE is sorted. Lack of sorting has nothing to do with preserving order. > -- > Sent from my phone. Please excuse my brevity. > > On March 8, 2017 8:55:06 AM PST, Dimitri Liakhovitski <dimitri.liakhovitski at gmail.com> wrote: >>Thank you. I was just curious what sort=FALSE had no impact. >>Wondering what it is there for then... >> >>On Wed, Mar 8, 2017 at 11:43 AM, Jeff Newmiller >><jdnewmil at dcn.davis.ca.us> wrote: >>> Merging is not necessarily an order-preserving operation, but sorting >>can make the operation more efficient. The sort=TRUE argument forces >>the result to be sorted, but sort=FALSE is in not a promise that order >>will be preserved. (I think the imperfect sorting occurs when there are >>multiple keys but am not sure.) You can add columns to the input data >>that let you restore some semblance of the original ordering afterward, >>or you can roll your own possibly-less-efficient merge using match and >>indexing: >>> >>> info[ match( grades2$grade, info$grade ), ] >>> -- >>> Sent from my phone. Please excuse my brevity. >>> >>> On March 8, 2017 8:07:27 AM PST, Dimitri Liakhovitski >><dimitri.liakhovitski at gmail.com> wrote: >>>>Hello! >>>>I have a vector 'grades' and a data frame 'info': >>>> >>>>grades2 <- data.frame(grade = c(1,2,2,3,1)) >>>>info <- data.frame( >>>> grade = 3:1, >>>> desc = c("Excellent", "Good", "Poor"), >>>> fail = c(F, F, T) >>>>) >>>> >>>>I want to get the info for all grades I have in info: >>>> >>>>This solution resorts everything in the order of column 'grade': >>>>merge(grades2, info, by = "grade", all.x = T, all.y = F) >>>> >>>>Could you please explain why this solution also resorts - despite >>sort >>>>= FALSE? >>>>merge(grades2, info, by = "grade", all.x = T, all.y = F, sort >>FALSE) >>>> >>>>Thanks a lot!-- Dimitri Liakhovitski