SAP Community - Machine Learning

DAR-GET bearer token - JSON error

2022-10-03T17:10:08+02:00

Tutorials: https://developers.sap.com/tutorials/cp-aibus-dar-setup-postman.html
--------------------------
Hi,

I am getting an unexpected token error while setting the bearer token. Using the api calls as-is imported from github. Has anyone faced this error ? Its the first step, i have been using DAR requests for the past 2 weeks now, never faced this issue

here is the body ..

Example regarding classification and clustering Type Machine learning in SAP PaPM? Thanks in advance

2022-11-09T07:29:10+01:00

Hi all,

Can someone please help me with an example regarding classification and clustering Type Machine learning in SAP PaPM?

Thanks in advance!

How can I use pdf invoice for retraining the DOX Service using API call in SAP IRPA

2022-11-28T11:30:13+01:00

Dear Team,

We have a scenario where we are using Document Information Extraction APIs to extract data(values, confidence and coordinates) from Invoice. Now we want to use DOX API retraining with same invoice with correct information using SAP IRPA. How can we retrain the DOX using API calls in SAP IRPA with correct Invoice which is having Accurate Data?

Please provide any blogs or any other document with proper information through which we can retrain the Document Information Extraction Service for better results in next execution. any update on this will be helpful.

Thanks.

------------------------------------------------------------------------------------------------------------------------------------------------
Learn more about the SAP Support user and program here.

How does artificial intelligence function?

2022-12-06T07:04:52+01:00

Artificial intelligence (AI) is the study of how robots replicate or mimic human intelligence, so clarify briefly how it functions and enable by several technologies.

No Scriptserver available - hana on cloud while running python program with hana_ml library

2023-04-27T14:20:45+02:00

Hi All,

I am facing below issue while trying to run a NaiveBayes ML model of hana_ml library from python program.

Issue : No ScriptServer available.

I am using Trial version of SAP BTP and HANA on Cloud.

Below is my Input Table.

Below is ML Model .

...

In above screenshot, it says "No ScriptServer available".

Please help me how to add/enable ScriptServer for hana on cloud.

How to directly call the API of S4 hana database in BTP Process automation?

2023-05-18T10:50:55+02:00

How can BTP Build Process Automation directly call the API of S4 PERSON_NAME database and how should it be implemented? Thank you very much.

Issue with consuming ML Scenario in SAP DI

2023-06-09T15:50:33+02:00

Hi,

I want to perform complete ML scenario based on SAP Academy "Financial Fraud Detection with SAP DI Cloud" (RandomForestClassifier from sklearn package is used there). In that Academy last step is model training, and I also want to test this model with new data. So, for next steps I have used SAP blog about creating ML-scenario in SAP DI.

Whenever I send request from Postman I get response ""Error message": "An error occurred: y should be a 1d array, got an array of shape () instead." It doesn't matter if I send values from Postman or use constant values in the script, the error remains the same. If I try to predict result with the same constant values in Jupiter notebook in SAP DI everything is OK.

What do I need to do to make it work?

It is the only change in script comparing with the blog mentioned before:

list_data = [21, 1, 3534, 1853, 5387, 18,  1184, -2350]

i_data = np.asarray(list_data)
input_data = np.reshape(i_data, (1, -1))

y_pred = model.predict(input_data)

Apply Machine Learning on Incident data set in excel format

2023-06-16T19:07:16+02:00

Hello All,

I have been given an Incident data set to provide insights out of the data. Also provide forecasts and predictions.

Please provide your suggestions to proceed on this do we have any prebuild service available.

Thanks,

For make a Machine Learning Model for Análisis Predictive in Costs Area ?

2023-06-18T18:24:30+02:00

What is better SAP DI or SAP SAC ?

We take costs from SAP using 10 Years Aprox.

How we can test trial for SAP DI ?

Machine Learning Extension in SAP HANA XSA (SOLMAN)

2023-08-11T09:10:45+02:00

We have checked below blog and it is mentioned about machine learning extension deployment in SAP HAAN XSA option to find anamoly detection of alerts.

Please help in sharing details/SAP Note/Document on how to perform machine learning extension deployment in SAP PERSON_NAME XSA of solution manager DB to detect anamoly alerts.

https://blogs.sap.com/2021/01/22/machine-learning-services-for-hybrid-operations-implementation-for-sap-solution-manager-7.2-and-sap-cloud-platform-sap-hana-xsa-part-2/

S/4 HANA and Python Integration

2023-09-06T17:35:06+02:00

Is it allowed to create, and update tables in S/4 hana database using python scripts from outside S/4 ?

Automatically Calculating tickets

2023-10-12T22:13:28+02:00

Hi SAP communicaty,

I want to know if SAP is able to automatically calculate weights onto loading tickets.

for eg. i have a van with a 3.5 ton max wt limit. - the van will be weighed before it is loaded with products on. I want to use SAP to be able to automatically calculate the amount of product it can carry based on its current weighed amount - max permissable wt. limit.

if anyone could gimme a clue on if SAP can do this and how, it would be much appreciated.

How to set privileges for HANA Machine Learning in Cloud Foundry ?

2023-10-13T08:17:49+02:00

Hi,

We need to use HANA_ML python package in HANA database in a Cloud Foundry App in BTP. While the code works perfect when running it on a jupyter notebook, it gives unsufficient privilege errors when calling it from within a cloud foundry deployment. (e.g. the ataframe.create_dataframe_from_pandas method fails).

We use the same code and user/password combination in both cases.

Any help appreciated on how to set authorization or how to allow deletion of tables from the CF deployment.

Regards

Marcus

Queries on SAP LLM and Joule?

2023-10-18T21:19:36+02:00

Hello Experts,

We are looking to enable Joule. However before doing so, we want to understand more about following question. Please share your thoughts on these questions.

1. Which LLM model is SAP using? Is it homegrown or Other models like GPT, BERT or LLAMA?

2. How is SAP addressing the hallucination issues?

3. Any other terms and condition that SAP want Client to accept in order to use Gen AI/Joule?
4. Is same LLM used in various modules or different LLM used ? For example for Job description one LLM, For SF chatboat another LLM model?
5. What are the update cadence for the SAP LLM Product like schedule release or Adhoc release?

6 .Where does this LLM resides and size of custom SAP LLM?

7. Once Joule is enabled, can we turn On/Off depending on modules?

How to use HANA-ML python package in a HANA HDI Container and how to assign correct priveledges?

2023-10-26T09:39:52+02:00

Hi ,

we are struggling with the use of an HDI container for machine learning with the python hana-ml package in Python.

In a previous question https://answers.sap.com/questions/13984833/how-to-set-privileges-for-hana-machine-learning-in.html?childToView=13984942 a topic on missing priviledges was raised. The answer in the end was to use the user/pw combination from the sharedaccesskey in the service definition. However, when using this user it needs to be hard coded. How can we assign priviledges "CREATE ANY" to the user which can be read via the python service object in python ? This is a different user. Also in the documentation it is not recommended to give CREATE ANY to a user in HDI container.

How can this issue be solved ?

Or is machine learning with python in general not allowed / recommended in a DHDI container ? That would however significantly reduce the value of the python approach.

Regards

Marcus

What transport to use from SAP Learning QA to PROD?

2023-11-10T12:49:45+01:00

Hello,

I hope you are fine,

Is it possible to enter all the elements in SAP Learning QA and then take them to Productive? Is there any transportation?

Thank you so much,

Kind regards

Integrate SAP Cash Application Cloud application with S/4 HANA 2022

2024-01-12T23:19:12+01:00

Hello,

I want to Integrate SAP Cash Application Cloud application with S/4 HANA 2022. Please share any Standard SAP Link/SAP Note/Standard SAP Guide which contains technical details to do the same.

Thanks and Regards

SAP HANA PAL Search table error: "Column doesn't exist"

2024-01-29T09:53:57.973000+01:00

Hello Community! 😄

When I was working with SAP HANA Predictive Analysis Library (SAP HANA 2.0 SPS05), I noticed an error in the procedure _SYS_AFL.PAL_LINEAR_REGRESSION_PREDICT. I’m wondering if any of you also got it.

According to the documentation columns of the parameter table are: „PARAM_NAME”, „INT_VALUE”, „DOUBLE_VALUE” and „STRING_VALUE”. Indeed, a table with such columns usually works and results are calculated correctly:

However, sometimes I get an error "Column ':PARAM_NAME' does not exist":

The same error is raised when the procedure is called with Python hana_ml library, but the names of columns that “do not exist” are different (without colons):

ERROR:hana_ml.algorithms.pal.linear_model:(256, 'sql processing error: "KDYMEK"."(DO statement)": line 13 col 1 (at pos 1233): search table error: _SYS_AFL.AFLPAL:FORECASTWITHLR_ANY: [102] (range 1) TypeMismatch exception: Column \'PARAM_NAME\' does not exist')
Traceback (most recent call last):
File "C:\Users\kdymek\AppData\Local\Programs\Python\Python310\lib\site-packages\hana_ml\algorithms\pal\linear_model.py", line 1069, in predict
self._call_pal_auto(conn,
File "C:\Users\kdymek\AppData\Local\Programs\Python\Python310\lib\site-packages\hana_ml\algorithms\pal\pal_base.py", line 859, in _call_pal_auto
self.execute_statement = call_pal_auto_with_hint(conn_context,
File "C:\Users\kdymek\AppData\Local\Programs\Python\Python310\lib\site-packages\hana_ml\algorithms\pal\pal_base.py", line 1365, in call_pal_auto_with_hint
if try_exec(cur, sql, conn):
File "C:\Users\kdymek\AppData\Local\Programs\Python\Python310\lib\site-packages\hana_ml\algorithms\pal\pal_base.py", line 1323, in try_exec
cur.execute(sql)
hdbcli.dbapi.Error: (256, 'sql processing error: "KDYMEK"."(DO statement)": line 13 col 1 (at pos 1233): search table error: _SYS_AFL.AFLPAL:FORECASTWITHLR_ANY: [102] (range 1) TypeMismatch exception: Column \'PARAM_NAME\' does not exist')

Workaround:

renaming columns of the parameter table: ":PARAM_NAME, ":INT_VALUE", ":DOUBLE_VALUE", ":STRING_VALUE" (with colons) – doesn’t work in Python,
or running the code once again as the error isn’t raised every time.

Do you know what causes this problem and how I can avoid it?

FedML SAP Datasphere & Databricks

2024-02-07T17:32:08.517000+01:00

Hello everyone,

I am interested in applying FedML to Datasphere data in Databricks and subsequently deploying the inference results back into Datasphere, while also deploying the Machine Learning Model to SAP Kyma. Before proceeding, I have several technical questions that I would like clarified:

1) When I call an SAP Datasphere View for consumption in Databricks using SQL, does the data remain solely in SAP Datasphere, or is it also replicated into Databricks via the "select" command?

2) Is the configuration for the DBConnection between SAP Datasphere and Databricks the same as the connection setup between Databricks and SAP Kyma?

3) Can inference be performed on the deployed model within SAP Kyma?

4) Apart from Azure, Amazon, and Google, are there any other hyperscalers supported by Databricks that I might be overlooking?

I am looking forward to your responses. Thank you in advance.

Best regards from Germany.

Re: Using PAL Procedures embeded in SQL Script (anonymous...

2024-03-01T12:55:38.026000+01:00

Hi Dirk,

based on your PAL procedure example, here is a simple example using anonymous blocks, you don't have to declare the structure of the PAL procedure output:

do
begin
tv_data = SELECT * FROM PAL_MULTIVARSTAT_DATA_TBL;
tv_parms = SELECT * FROM #PAL_PARAMETER_TBL;
CALL "_SYS_AFL"."PAL_MULTIVARIATE_ANALYSIS"(:tv_data, :tv_parms, tv_out);
SELECT * FROM :tv_out;
END;

The constraint if structures require to be declared or not comes from SQL Script.

Please note, usage of the Python Machine Learning client might be helpful, as you can script in Python and capture the ad-hoc generated SQL for the PAL procedure or even utilize methods to generate procedure code / even design-time artifacts. Just as another helping note.

I hope this is helpful, otherwise please share more details.

Best regards,

Christoph