Error 314: numeric overflow: search table error: 6499 error executing physical plan: exception 6499

[Former Member]

Hello all, I'm having several troubles with PA recently and one of then, is the error below.

It happens after I execute the APRIORI algorithm, and try to visualize the results. So then appears this window:

Does anyone knows what might be happening?

Regards!

[Former Member]

Hello all,

I keep searching any solution the past days, and I've done the following:

I changed the agregation of the measure collumn (NUMERONOTAVENDA), it was SUM and I changed to COUNT.

So then I ran the APRIORI and it worked, but didn't show any results. SO then I reduced the Confidence from it (80% -> 75%) and re-executed, but it gave me the following error: Registration finished with errors, see indexserver trace.

So than I put the confidence back to 80%, re-executed and gave the same error!

Than I closed PA, and make all that steps again, and I noticed that Apriori only works the first time I open PA, but never brings me any result.

Does anybody came across this error before?

Regards!

[Former Member]

Hey Jurgen,

As Andreas mentioned, you are close now. Your PA is working fine with HANA.The error is of HANA and nothing to do with Apriori or SAP PA.

You get that error when you hit the upper limit of the datatype on column "NUMERONOTAVENDA" since you are using SUM aggregation.

Alter the table column datatype to higher (say DOUBLE), or you can try another workaround with datatype conversion function (TO_DOUBLE)

eg: convert the column type to double using TO_DOUBLE(NUMERONOTAVENDA) while creating the analytic view

See if it works and let me know.

[Former Member]

Hi Bimal, do you know where I can learn the differences between the kind of aggregations? (SUM, MIN, MAX, COUNT)

DO I need to change the datatype directly at the table (inside my schema), or at the analytics view I created?

Regards! and thanks for you help!!

[Former Member]

I don't know if there exists such a document that will tell you difference between different kind of aggregations but I can explain with an example , but the blog below should help you

If you have a dataset like below:

Product	Customer	Transaction Id	Value
P1	C1	1001	300
P1	C1	1002	400
P1	C1	1003	500
P1	C2	1004	600
P1	C2	1005	100
P2	C1	1006	750
P2	C2	1007	800

If you create a view with only Product and Customer, the data for different aggregations appears as below:

SUM

P1	C1	1200
P1	C2	700
P2	C1	750
P2	C2	800

MIN

P1	C1	300
P1	C2	100
P2	C1	750
P2	C2	800

MAX

P1	C1	500
P1	C2	600
P2	C1	750
P2	C2	800

COUNT

P1	C1	3
P1	C2	2
P2	C1	1
P2	C2	1

Hope you get it.

Regarding change in data type, do it at the table and view both to avoid any issue (if space is not a constraint for you)

[Former Member]

Hi Bimal, thank you so much for your explanation about the different types of aggregations!

I'm thinking about using the following SQL Command:

ALTER TABLE "ABC"."INTEGER" ALTER (COL_NUMERONOTAVENDA DOUBLE);

Where the table name is: ABC, the format of the collumn I want to change is INTEGER

The collumn I want to change is named as NUMERONOTAVENDA

and the new format is DOUBLE

Am I doing it right?

Can I execute this command?

Regards!

[Former Member]

The statement looks correct to me. This will help you.

SAP HANA Reference: ALTER TABLE

First I would say, change the column type in view. Ideally it should work.

If it does not than goto Table and modify it.

Try both and see which one works for you.

[Former Member]

Hi Bimal, tha SQL Command that I created doesn't works!

I changed some parameters, and it keeps failling.

Can you please explain me how do I change the datatype directly in the view?

Do I change at the analytic view I've created, or in the original table?

Regards

[AndreasForster]

Hi Jürgen,

I think you are almost there.

It sounds as if the Apriori execution succeeds and only when going to see the results you get this error? That error is probably related to the column type. Before changing date types, you could quickly try these two things to narrow the problem down

- Put a filter component before the Apriori component. Apply a row filter in it to restrict the number of rows. Use something very small, maybe just 1000 rows.  If this works, you are sure the error is related to the larger number of rows/values

- If the above is working, then remove the filter and add a HANA Writer component after Apriori. Now executed the analysis and do not display the results in SAP PA. Instead see if the HANA table was filled with the Apriori results.

Greetings

Andreas

[Former Member]

Hi Andreas, I'm trying to do the the steps that you recommended me, and I'm with a few doubts about it:

First I tried to add a filter, so then I've done the following:

I noticed that the only one that worked was the collumn named "NUMERONOTAVENDA" so then I realized that it was because that column is the only measure that I have (I've created an analytic view, and put that collumn as a measure) so then I put it (on HANA) as a SUM (I don't understand the difference betwenn the options I can select over there (SUM, MIN, MAX and COUNT)), so then I put it as SUM.

Ok, comming back to PA, I clicked in execute, and then it worked, and when I clicked to see the results, it gave me the same error:

Did I do anything wrong, or that is a better way to do that?

ok, so then tried to do the second step you suggested me: put on and executed the APRIORI algorithm, it worked, but I didnt changed to the results view. and after that, I used the hana writer:

and clicked ok, so then I clicked again in execute, and got this error:

Then I went to HANA Studio, the right schema, and saw the table over there, but when I try to Preview Data, it opens a blank table!

Again, did I miss anything, or am I doing anything wrong?

regards!