Strange Error in PA

[Former Member]

Hi,

I am reading this tutorial : https://github.com/AndreasForster/Predictive/raw/master/SAP_PA_Automated_Tutorial_DataManager.pdf

I am now in page 15 and clicked on "View Data"  to see the content of Entity, but I got this error:

2016-04-01 14:34:39    Error    Message from ODBC: SQLExecDirect: [S1000][SAP AG][LIBODBCHDB DLL][HDBODBC] General error;260 invalid column name: RKARIMI_DEVUSERCLIENT.CLIENT_ID: line 1 col 38 (at pos 37)

The column CLIENT_ID exist in table and was already selected as the Id field. Moreover, in the tutorial there is DATE_BIRTH column but it does not exist in the data set that I downloaded from Hands-On Tutorial SAP Predictive Analytics, Aut... | SCN.

.Thanks,

Rasoul

[achab · Accepted solution]

Hi Rasoul, looping Former Member. Thanks & regards, Antoine

[Former Member · Accepted solution]

Thanks Antonie! You mean I should contact him or he will contact me?

[AndreasForster · Accepted solution]

Hi Rasoul, Strange thing, as it was working for other people. So I wonder whether it is something specific to your environment.

Which version of SAP PA and HANA are you using?

Greetings

Andreas

[Former Member · Accepted solution]

Hi Andreas,

I am using Version: 2.5.0 and Version: 2.1.15

Thanks,

Rasoul

[Former Member · Accepted solution]

Sorry I means PA Version: 2.5.0 and HANA Version: 2.1.15

[achab · Accepted solution]

Hi Rasoul, the version you give us is the SAP HANA Studio one, we need the one of the SAP HANA system; screenshot to follow.

Thanks & regards

Antoine

[Former Member · Accepted solution]

There is it:

[achab · Accepted solution]

Revision 102.01, which is supported by PA 2.5.

Thanks

Antoine

[AndreasForster · Accepted solution]

Thank you Rasoul and Antoine.

I have just imported the data myself and I do see the DATE_BIRTH column in the CLIENT table.

Can you please try importing again. Is there anything unusual, any warnings?

[achab · Accepted solution]

Hi Andreas, hi Rasoul,

I have the same result on my end FYI.

Cheers,

Antoine

[Former Member · Accepted solution]

Thanks Antonie and Andreas for your effort.

I just noticed that I had downloaded the data from  http://lisp.vse.cz/pkdd99/DATA/data_berka.zip

I could not import the data that I downloaded from Hands-On Tutorial SAP Predictive Analytics, Aut... | SCN

because it is not a text file and I don't know how to import it! To import a text file into HANA first I open it in Excel and then create a xlst file and then import it using import wizard in HANA Studio.

So it seems that something is wrong in the file that I downloaded from the first link because the DATE_BIRTH does not exist and the client_id is problematic. Could you please tell me how to import the data from the second link? By the way, is the type of client_id column important? I tried INTEGER and DECIMAL but both failed. Is it also important to define client_id as a primary key? I added this constraint but it did not help!

Thanks,

Rasoul

[achab · Accepted solution]

Hi,

Look in Andreas document SAP_PA_Automated_Tutorial_DataManager.pdf starting from page 64.

Cheers

Antoine

[Former Member · Accepted solution]

Hi,

I just fixed this issue. The problem was that the column names in the table are in small letters but PA expects them in big letters!

In the end of "Hands-On Tutorial SAP Predictive Analytics, Automated Mode: Data Manager", it is mentioned: "It is preferred to use capital letters for all table names and column names, ie use CLIENT_ID instead of ClientID". I think this sentence should be changed in order to make it mandatory not optional. Moreover, in the original data files (http://lisp.vse.cz/pkdd99/DATA/data_berka.zip ),

the name of columns in the text file should be in big letters in order to avoid this problem automatically!

Best,

Rasoul

[achab · Accepted solution]

Hi,

Good to know it is resolved - I had noticed the same difference (but did not think this was causing the error). In the link you mention the files are not provided by SAP, we have no control over this. Could you please be so kind to mark the question as Answered?

Thanks

Antoine

[AndreasForster · Accepted solution]

Hi Rasoul,

The file data_berka.zip was kindly shared by someone organizing an analytical competition. This data is not coming from SAP, as Antoine rightly mentions.

The tutorial SAP_PA_Automated_Tutorial_DataManager.pdf makes use of that data, which requires a number of transformations though. To follow the tutorial it would be easiest for you to download the transformed dataset. You can import it in a few clicks. Please see page 64 in SAP_PA_Automated_Tutorial_DataManager.pdf. So instead of using data_berka.zip I recommend using DataManagerTutorialData.zip. Any transformations made to the original data is described in the tutorial.

Apparently it is possible to use lowercase as table/field names by use of some clever connectivity settings. Hence it's not declared manadtory.

Many Greetings

Andreas

[Former Member · Accepted solution]

Hi,

I have imported the data using "Catalog Object" Wizard and have progress until page 37 of the tutorial. Now I am creating a model by hitting "Generate " button. Then I get this error:

Error: All the input variables are constant. No model can be built.

This is strange because like the tutorial I have selected "Gender" and "DISTRIC_ID" as input variables and these variables are not constant when I view the data (attachment 2). Just for test, I added more columns as input variable but it did not help.

Any idea?

Thanks,

Rasoul

[Former Member · Accepted solution]

... I also send you the log file, it may be helpful.

[Former Member · Accepted solution]

Sorry, my bad, I just figured it out that the variable "BuysCreditCardInQuarterAfterTimeStamp” was set to "NoExist" but "Exist" is correct. I am working on it!

[achab · Accepted solution]

Hi, in general, if you have new questions, can you please create new posts? this will ease the follow-up. Thanks in advance, Antoine