on 08-29-2015 10:46 AM
Hi
I am trying to run auto classification algorithm with the data set having around 4,50,000 rows with 50 columns. I am using 8 GB machine.
Let me list down the error i faced up until now for the same dataset:
BUG1 : First time when I tried running the model with around 30 columns as input and 1 Target Variable. Below attached error popped up. Despite the fact I have allocated 4GB virtual memory for the tool using .ini file
[Sybase][ODBC Driver][Sybase IQ]You have run out of space in IQ_SYSTEM_MAIN DBSpace.
-- (s_blockmap.cxx 4566)
undefined
However after the multiple restarts of the tool and doing trial runs, the model run successfully took about 12 mins.
I got the output, did some visual analysis and created a story board and saved the document.
BUG2: Ok after sometime when I tried reopening the document, a new error pops up:
magically the story board was there but in the predict tab I can't see the model anymore.Blank predict tab, no algorithm mapping nothing. It happened earlier also I posted the same issue on scn previously but no concrete solution yet .
BUG3: Ok I thought may be data is too large, I deleted few junk columns and the dataset size got reduce to 173 MB from 234 MB. And I applied the Auto Classification algorithm. The model ran successfully, and now when I am trying to apply Model Statistics component to the model, a new error pops up
and this above error didn't came first time when I saved the document earlier which got corrupted.
I really don't get it whats the issue. My more focus and time is taken away on fixing such issues, rather than working on the algorithm.
Does anyone has any suggestion on how to handle such issues or why its happening like that?
Regards
Ranajay
Hi Ranajay,
Could you tell me how you managed to extend the space for the IQ server?
I've encountered the same error with fewer data ( 1MM rows & 6 columns).
Regards,
Dan
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Dan,
I would recommend that you upgrade first to SAP Predictive Analytics 2.4 and see if you still face the problem before you tweak the .ini file.
In case you need further help, can you please create a new discussion with all the details on the problem that you are facing?
Thanks & regards,
Antoine
At this stage, I do not have 100% certainty that the problem is caused by a memory problem (despite the fact it looks like it is).
Hence my recommendation to open a dedicated discussion on this, Dan.
It's also good to know that SAP support would focus help on software that is operated "out of the box".
And chances are we will need to go through the support channel on this one, in case it is reproduced on SAP PA 2.4.
Best regards
Antoine
Hi Antoin,
It seems the issue persists, with the file as it is or with the memory extended as per Ranajay's comment.
I even tried extending the values to huge amounts:
I'm getting this error by using standard components (I haven't even reached the point where I can deploy my custom component):
What do you suggest would be the best thing to do next? Open a new discussion (as this is related to 2.4) or open a support ticket?
Just an FYI, I'm using the Trial version of SAP PA so I am not sure if I am able to open a support ticket for it...
Regards,
Dan
Hi Ranajay
Please log a support ticket with SAP support so that we keep track of these issues not only via SCN discussions.
Thanks & regards
Antoine
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
86 | |
23 | |
11 | |
9 | |
8 | |
5 | |
5 | |
5 | |
5 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.