# Get first row from a group **URL:** \ **Category:** Community Support\ **Tags:** kdb-and-q\ **Created:** [January 5, 2016, 12:55am UTC](https://forum.kx.com/t/get-first-row-from-a-group/10718 "2016-01-05T00:55:00Z")\ **Posts on this page:** 8\ **Page:** 1
**Author:** ![carfield11](https://avatars.discourse-cdn.com/v4/letter/c/ec9cab/32.png) [@carfield11](https://forum.kx.com/u/carfield11)\ **Post date:** [January 5, 2016, 12:55am UTC](https://forum.kx.com/t/get-first-row-from-a-group/10718/1 "2016-01-05T00:55:00Z")
Hi, if I like to select the last row from a group, I can just do select by xxx from table if there similar syntax to select the first row? Other than using “first” for all elements? ---
**Author:** ![kuentang](https://avatars.discourse-cdn.com/v4/letter/k/bc8723/32.png) [@kuentang](https://forum.kx.com/u/kuentang)\ **Post date:** [January 5, 2016, 8:38am UTC](https://forum.kx.com/t/get-first-row-from-a-group/10718/2 "2016-01-05T08:38:00Z")
Something like this?\\ \ \ select by xxx from reverse table\\ \ \ \ \ Kim\\ \ \ **Von:** [personal-kdbplus@googlegroups.com](mailto:personal-kdbplus@googlegroups.com) [[mailto:personal-kdbplus@googlegroups.com](mailto:personal-kdbplus@googlegroups.com)] **Im Auftrag von** Carfield Yim **Gesendet:** Dienstag, 5. Januar 2016 01:55 **An:** Kdb+ Personal Developers **Betreff:** [personal kdb+] Get first row from a group\\ \ \ Hi, if I like to select the last row from a group, I can just do select by xxx from table if there similar syntax to select the first row? Other than using “first” for all elements? \\ – Submitted via Google Groups ---
**Author:** ![carfield1](https://avatars.discourse-cdn.com/v4/letter/c/ba8739/32.png) [@carfield1](https://forum.kx.com/u/carfield1)\ **Post date:** [January 6, 2016, 4:15pm UTC](https://forum.kx.com/t/get-first-row-from-a-group/10718/3 "2016-01-06T16:15:00Z")
Thanks, yes, but if reverse a whole table expensive? ---
**Author:** ![User7](https://avatars.discourse-cdn.com/v4/letter/u/7ea924/32.png) [@User7](https://forum.kx.com/u/User7)\ **Post date:** [January 6, 2016, 5:11pm UTC](https://forum.kx.com/t/get-first-row-from-a-group/10718/4 "2016-01-06T17:11:00Z")
Hi Carfield, You can use this: `xxx xkey table value exec first i by xxx from table It’s a little more verbose, but will avoid the need to name columns and also the performance impact of a table reverse. Regards **Cathal Deehan** Financial Software Developer AQUAQ Analytics ---
**Author:** ![mkrmkr1](https://avatars.discourse-cdn.com/v4/letter/m/ba9def/32.png) [@mkrmkr1](https://forum.kx.com/u/mkrmkr1)\ **Post date:** [January 6, 2016, 9:43pm UTC](https://forum.kx.com/t/get-first-row-from-a-group/10718/5 "2016-01-06T21:43:00Z")
you don’t need to reverse the table. a function solution will work best. here’s an example: q)t:(a:1 1 1 2 2 2;b:`a`b`c`d`e`f;c:10 20 30 40 50 60) q)t a b c ------ 1 a 10 1 b 20 1 c 30 2 d 40 2 e 50 2 f 60 q)select by a from t a| b c -| ---- 1| c 30 2| f 60 q)/is the same as q)select last b,last c by a from t a| b c -| ---- 1| c 30 2| f 60 q)/so for the first row q)select first b,first c by a from t a| b c -| ---- 1| a 10 2| d 40 q)/but for lots of columns that’s tedious so q)?[`t;();enlist[`a]!enlist`a;(cols[`t]except`a)!(first,/:cols[`t]except`a)] a| b c -| ---- 1| a 10 2| d 40 q)/which you could make into a little utility q)firstrow:{[t;g]?[t;();enlist[g]!enlist g;(cols[t]except g)!(first,/:cols[t]except g)]} q)firstrow[t;`a] a| b c -| ---- 1| a 10 2| d 40 q) ---
**Author:** ![TerryLynch1](https://avatars.discourse-cdn.com/v4/letter/t/9e8a1a/32.png) [@TerryLynch1](https://forum.kx.com/u/TerryLynch1)\ **Post date:** [January 6, 2016, 10:04pm UTC](https://forum.kx.com/t/get-first-row-from-a-group/10718/6 "2016-01-06T22:04:00Z")
What’s wrong with select from table where i=(first;i) fby keycol Just have to be careful of using “i” with partitioned tables. Also, the above can be easily changed to grab last, or first 2 rows per key etc. And can be used for multiple keys select from table where i=(first;i) fby (keycol1;keycol2) ---
**Author:** ![carfield11](https://avatars.discourse-cdn.com/v4/letter/c/ec9cab/32.png) [@carfield11](https://forum.kx.com/u/carfield11)\ **Post date:** [January 7, 2016, 2:56pm UTC](https://forum.kx.com/t/get-first-row-from-a-group/10718/7 "2016-01-07T14:56:00Z")
Thanks a lot everyone! ---
**Author:** ![Aaron\_Davies](https://avatars.discourse-cdn.com/v4/letter/a/8e7dd6/32.png) [@Aaron\_Davies](https://forum.kx.com/u/Aaron_Davies)\ **Post date:** [January 11, 2016, 12:42am UTC](https://forum.kx.com/t/get-first-row-from-a-group/10718/8 "2016-01-11T00:42:00Z")
note that this solution is the only one that works on partitioned tablescan be shortened and generalized to work with multi-column group as{?[x;();y!y,:();c!first,/:c:(cols x)except y]}(just to throw another one out there, there’s also {first each’xxgroup y} — still won’t with partitioned tables though, and it’sprobably the most expensive in both time and space)On Wed, Jan 6, 2016 at 4:43 PM, mkrmkr wrote:\> you don’t need to reverse the table. a function solution will work best.\> here’s an example:\>\> q)t:(a:1 1 1 2 2 2;b:`a`b`c`d`e`f;c:10 20 30 40 50 60)\> q)t\> a b c\> ------\> 1 a 10\> 1 b 20\> 1 c 30\> 2 d 40\> 2 e 50\> 2 f 60\> q)select by a from t\> a| b c\> -| ----\> 1| c 30\> 2| f 60\> q)/is the same as\> q)select last b,last c by a from t\> a| b c\> -| ----\> 1| c 30\> 2| f 60\> q)/so for the first row\> q)select first b,first c by a from t\> a| b c\> -| ----\> 1| a 10\> 2| d 40\> q)/but for lots of columns that’s tedious so\> q)?[`t;();enlist[`a]!enlist`a;(cols[`t]except`a)!(first,/:cols[`t]except`a)]> a| b c> -| ----> 1| a 10> 2| d 40> q)/which you could make into a little utility> q)firstrow:{[t;g]?[t;();enlist[g]!enlist g;(cols[t]except> g)!(first,/:cols[t]except g)]}> q)firstrow[t;`a]\> a| b c\> -| ----\> 1| a 10\> 2| d 40\> q)\>\>\>\>\> On Wednesday, January 6, 2016 at 11:15:43 AM UTC-5, Carfield Yim wrote:\>\>\>\> Thanks, yes, but if reverse a whole table expensive?\>\>\>\> On Tue, Jan 5, 2016 at 4:38 PM, Kim Tang \ wrote:\>\>\>\>\>\> Something like this?\>\>\>\>\>\>\>\>\>\>\>\> select by xxx from reverse table\>\>\>\>\>\>\>\>\>\>\>\>\>\>\>\>\>\> Kim\>\>\>\>\>\>\>\>\>\>\>\> Von: [personal...@googlegroups.com](mailto:personal...@googlegroups.com) [[mailto:personal...@googlegroups.com](mailto:personal...@googlegroups.com)]\>\>\> Im Auftrag von Carfield Yim\>\>\> Gesendet: Dienstag, 5. Januar 2016 01:55\>\>\> An: Kdb+ Personal Developers\>\>\> Betreff: [personal kdb+] Get first row from a group\>\>\>\>\>\>\>\>\>\>\>\> Hi, if I like to select the last row from a group, I can just do\>\>\>\>\>\> select by xxx from table\>\>\>\>\>\> if there similar syntax to select the first row? Other than using “first”\>\>\> for all elements?\>\>\>\>\>\> –\>\>\> Submitted via Google Groups\