Hello All, I know there is a similar topic for this but I can’t seem to get this to work, wondering if someone can help
I have a table similar to
t:([]sym:$();val:“I”$());t insert (A;500);t insert (A;1000);t insert (A;1500);t insert (A;-1000);t insert (A;-1500);t insert (A;500);sym valA 500A 1000A 1500A -1000A -1500A 500A 500`
I’d like to have a running sum, but have it reset when signum changes, so would end up like
sym val val_sumsA 500 500A 1000 1500 A 1500 3000A -1000 -1000A -1500 -2500A 500 500A 500 1000
I think this should be straight forward but I can’t seem to get it, any help would be appreciated.
Thanks!
Hi Roni,
The following update statement will provide the running sum resetting when the signum changes,
update val_sums:{y+x*signum
Regards
Karan Patel
This will give you what you want:
update val_sums:sums val by sums differ signum val from t
q)tsym val---------A 500A 1000A 1500A -1000A -1500A 500A 500q) update val_sums:sums val by sums differ signum val from tsym val val_sums------------------A 500 500A 1000 1500A 1500 3000A -1000 -1000A -1500 -2500A 500 500A 500 1000
This is slightly quicker than my previous one:
update val_sums:sums val by sums differ val<0 from t
update val_sums:{y+x*z}[0;val;not differ signum val] from t
not differ signum val can also be used to reset the running sum
Regards
I can get crudely kind of close but its still wrong and I think there must be a better way
update val_sums:{[x;y;z]?[y;x;x+z]}'[val;differ signum val;prev val] by sym from t sym val val_sumsA 500 500A 1000 1500A 1500 2500 -> this is wrongA -1000 -1000A -1500 -2500A 500 500
User7
June 8, 2018, 3:24pm
7
Hi Roni,
How about something like this:
ungroup delete val from select val_sums:sums val by sym, sums differ signum val from t
Hi,
Think this should do it
q)update sums_val:raze sums each cut[;val]where differ signum val from t
sym val sums_val
------------------
A 500 500
A 1000 1500
A 1500 3000
A -1000 -1000
A -1500 -2500
A 500 500
Jason
ahh yea that works, nice, thanks!
This should work:
q)update val_sums:sums val by sums differ val<0 from t
sym val val_sums
A 500 500
A 1000 1500
A 1500 3000
A -1000 -1000
A -1500 -2500
A 500 500
perfect, yea this is great
q)update val_sums:sums val by sums differ val<0 from t
you guys are awesome
Here is yet another option which seems to be quicker but uses more memory:
update val_sums:raze sums each (where differ val<0) _ val from t
q)\ts:10000 update val_sums:sums val by sums differ val<0 from t
Attila
June 12, 2018, 12:09pm
13
for longer tables we can also do this<o:p></o:p>
q)T:100000#t<o:p></o:p>
q)\ts update val_sums:sums val by sums differ val<0 from T<o:p></o:p>
9 5365536 <o:p></o:p>
q)\ts update vals:raze sums each (where differ val<0) _ val from T<o:p></o:p>
4 3268352 <o:p></o:p>
q)\ts update val_sums:{s-0i^(s:sums x)i bin -1+i:sums differ x<0}val from T<o:p></o:p>
1 3147024 <o:p></o:p>
<o:p style=‘"background-color:’ rgba> </o:p>