SAP Community - Machine Learning

Custom Field not getting updated

2024-12-10T07:22:53.210000+01:00

Custom field ethnicity/Race is configured to sync from EC into LMS through the connector, checking the archives in connector schedulling, the field does come from but when you go to the Label on the profile in LMS it is not showing.

sfuser.extendeduser.field.mapping.data.COL_NUM3_VAL=OCCUPATIONAL_LEVEL
sfuser.extendeduser.field.mapping.data.COL_NUM4_VAL=CHALLENGE_STATUS
sfuser.extendeduser.field.mapping.data.COL_NUM5_VAL=ETHNICITY
sfuser.extendeduser.field.mapping.data.COL_NUM7_VAL=RISK_LOSS
sfuser.extendeduser.field.mapping.data.COL_NUM8_VAL=IMPACT_LOSS

@PL1

@Harish_Vatsa

SF-Dan

Missing Execution ID in SAP AI Launchpad After Connecting AWS S3 Bucket via Object Store Secret

2025-01-16T20:59:41.617000+01:00

I'm using SAP AI Launchpad. I created an Object Store Secret and connected it to my bucket in AWS, as it is a demo the S3 has admin role privilege in order tu access and has an access point, but in my dataset in SAP AI Launchpad, I don't see an execution ID. What can I do?

Sci-Kit learn library in SAP Datasphere Data Flow

2025-01-22T16:41:14.270000+01:00

Hi everyone,

I tried to add Python code into a Sap Datasphere Data Flow utilizing the sci-kit learn library. However I always get errors. Does anyone know if the library is available in the Dataflow Script Console or what alteratives I would have here?

Many thanks,

Moritz

Access event log, case attributes via API to perform ML based predictions

2025-02-25T15:05:56.437000+01:00

Hi Team,

I am looking for options to access Business objects, event log created within Process Intelligence. I would require to access this data to perform ML based predictions or employ other AI based applications.

The idea is to perform all these computations in external notebook and push this data back to Process Intelligence so that we can show predictions or any other applications.

Has anyone worked on this kind of setup, how to access data from Process Intelligence and push it back to the model. Any kind of API provided or Python based setup available, please advise.

@Process Intelligence @Pa_Vi Signavio

Joule Extensibility POC for SAP Learning

2025-02-26T06:32:43.310000+01:00

Hello everyone,
We're considering a proof of concept that involves utilizing the Joule Extensibility feature in relation to the SAP Learning platform (Learn SAP skills | SAP Learning). The idea is to implement a chatbot that enhances user experience by offering improved recommendations and roadmaps based on their search queries. I'm curious about the feasibility of this concept, given that Joule can currently be integrated only with SAP Line of Business products. Is it possible for us to proceed with this use case?

How to enhance Machine Learning for C4C v1

2025-03-10T07:59:07.312000+01:00

Hi - we have a new requirement where the ML for Service should only work for specific Ticket Type and Channel. I know we have Enhancement Spot for ML but not sure where to start.

Thanks!

Marc

Deployment is in Pending Status since 3 days in SAP AI launchpad - ML Deployment

2025-03-14T11:27:18.614000+01:00

Greetings Community Members!

Kindly help me with understanding on Deployment troubleshooting. I have deployed a vision model using Postman, it was working well then I made some changes in re-deployed again. It has been in pending status since last 3 days. Is there any ways to know Logs, debug, or troubleshoot deployment process ? in AI launchpad - ML( Deployment Section) Logs are very basic not providing more information, is there any way we can see more detail logs for the deployment ?

Thank you
Best Regards
Naim

SAP AI Core

# AI Launchpad

#SAP ML Deployments

Where to maintain ISLM Connection mapping in S/4HANA Public Cloud

2025-03-21T13:24:07.672000+01:00

I was following multiple blogs to understand what is ISLM. There was a place where they were doing configuration in ISLM Connection mapping in SPRO. How to do that in S/4HANA Public Cloud version.

Unable to create Intelligent Scenarios in S/4 Public Cloud

2025-03-27T07:31:14.382000+01:00

When I try to create a Intelligent Scenario in S/4 HANA Public Cloud, I am getting below error.
Unable to edit the field.

Questions about SAP Joule

2025-04-15T07:07:51.942000+02:00

Hello SAP Community,

We are planning to enable SAP Joule. But before we proceed, we wanted to understand the capabilities of Joule with respect to SAP Sales Cloud Version 2. I had a few questions regarding it, please read through them and provide your insights.

I understand that SAP Joule currently has 3 Interaction Patterns: Transactional, Navigational and Informational. But I would like to know more about the following:

1. What are the Business Objects supported in Joule with respect to SAP Sales Cloud Version 2?
2. What are the use-cases that are supported in Joule with respect to SAP Sales Cloud Version 2?
3. What are the current limitations of Joule?
4. Is there a way to customize navigation capabilities of Joule from SAP Sales Cloud Version 2 to SAP ECC?

If there are any online sources that I could reference for the questions above, please provide those to me as well.

Thanks in advance,
Karthik C M.

Detecting Handwritten vs. Digital Docs in SAP CAP + DOX

2025-05-09T08:46:00.939000+02:00

I'm working on a solution using SAP CAP (Cloud Application Programming) with Node.js, integrated with SAP Document Information Extraction (DOX). This project sends documents via API to the DOX service to extract relevant information from files such as invoices and receipts. My requirement here is to check whether the document is handwritten or computer-generated. This distinction could significantly enhance data accuracy and processing logic. If anyone has insights on integrating handwriting detection with SAP DOX or using pre-processing techniques for document classification, I’d love to connect and learn more.

The AI Adoption Journey: Machine Learning- A Deeper Dive!

2025-05-19T18:00:38.758000+02:00

Hello all!

Last blog, we skimmed over some of the Machine Learning features that SAP Fieldglass has but I’ve since gotten to experiment with it more and I must talk about it!
First things first. Machine Learning is a part of the base AI for Fieldglass, which means its available for anyone that has Fieldglass. It blows my mind that it’s not used more!

So, why would anyone want to use these features? Like the example I gave in the last blog, getting a large influx of resumes from various job postings can be a substantial burden on your administrative team.

How does Resume Assessment work to resolve this? AI assists in identifying the right job seekers by screening submitted candidates in how close of a match they are to the job description.
I remember when I was starting college, the advice was to put all the important information around the middle of the page, where a person’s thumb would be holding it. This was the idea they used to have about “efficiency” when it came to reviewing resumes. I'm sure we've all heard little tips and tricks like that- repurposing keywords in the resumes, placing them somewhere specific on the page, even hacks on fonts to keep the readers eyes on the page!
Now, we have AI that will comb through the candidate resumes to do a match on both skills and work experience.
Your business has more sway in this process than you might think! Buyers can prioritize the factors differently, depending on industry, job title or level of experience! A new field called Resume Assessment Weighting on the Contingent Type can be leveraged to indicate the level of importance the business deems appropriate.
This field contains four main categories that the buyer can set based on their preference to produce a banded score: Skills, Industry Experience, Work Experience and Qualifications.

The algorithm matches the skills on the resume The algorithm matches the skills listed on the resume against the application's skills library, analyzes the industry experience, work experience, and qualifications based on the preferences set using the Resume Assessment Weighting field, and produces that banded score. This scoring system uses a five-tier band: best, better, good, weak, no match) to indicate compatibility with the job posting description.
So why the banded score? This high-level score can let the administrative team know, immediately, who the most compatible candidates are for the Job Postings. With this feature in use, businesses saw that 61% of candidates hired were ranked “Best” by resume assessment.
ML doesn’t stop here with the requisition process. There’s also an experience assessment and skills extraction The Experience Assessment will offer a percentile score (1-100%) based on skills and related experience, as well as skills extraction, where AI combs through the submitted resume, and will visually highlight skills and qualifications making them easier to see (freeing up your thumbs and saving paper). This not only speeds up the reviewing, short listing and hiring process. Customers that enabled this feature saw an over all 15% reduction in workers closed with a “do not hire” reason.

These seemingly simple change in automation removes bottlenecks in the process in speeding up review to hire but also cleans up the integrity of the workforce hired overall.

Of course, there is all this information and so much more to play with in an interactive setting on the SAP Fieldglass Product Work Zone. They have an amazing AI Hub that I strongly encourage you to explore: SAP Fieldglass Product Success Work Zone - AI Hub

Also, please check out this great series by Fabian Caelers on Quick Win Features to learn some more valuable tips and tricks to make the most of your solution: SAP Fieldglass - Quick Win Features - Episode 2 - SAP Community

Join me next week, as I get into the obscure gem that is O*NET code recommendations in Fieldglass and why that matters!

SAP BDC Demo Environment

2025-05-21T10:07:19.202000+02:00

Hello All,

Since the BDC trial version is now available, I’d like to create a demo using it. Could you please guide me on how to enable or build the content? Also, is it possible to showcase a demo using standard content, considering that the source systems are not properly maintained? I’m looking for any way to effectively pitch it to clients. I’m exploring options to work exclusively within BDC AI/ML part. The practice source systems are not properly maintained, making it impractical to establish connections with them. However, intelligent applications or insight apps require both source and target systems to be connected in order to enable the content. Are there any materials or documents available that can guide the implementation of Databricks AI/ML on raw data within an SAP BDC environment?

DATA Science

2025-05-26T07:00:19.653000+02:00

How I can perform Machine Learning, Deep Learning and NLP programs on SAP Platform and in which SAP platform I should use like, BAS with Jupyter notebook or Databricks or something else.

Intelligent Scenario Management model training failed

2025-05-31T06:51:35.798000+02:00

Dear expert,

I'm facing authorization error for training existing intelligent scenario SUPLRDELIVPREDICT. I'm having SAP_BR_ADMINISTRATOR role but still not authorized to train the model.

Training failed : SUPLRDELIVPREDICT - model Suplr_Delvry_PredYou don't have any authorization to Create Model Version for Intelligent Scenario SUPLRDELIVPREDICT.

I tried to trace the missing auth but i dont see failed object in Authorization tracing.

Any idea what role is missing to train model in Intelligent Scenario?

@PritiDhingra

Regards,

Ben

SAP AI service running on-premise

2025-06-03T15:40:06.556000+02:00

Dear,

I am exploring to use SAP BTP AI service for client use cases, my problem is client is insisting to run AI use cases in client on-premise environment.

Is it possible to create and run AI solutions in on-premise environment?

thankyou

ebsor

Feature Engineering in SAP BTP: SAP HANA Cloud In-built ML vs. SAP Datasphere

2025-07-16T17:19:20.369000+02:00

Hello Team,

We are implementing a new predictive use case on SAP BTP. We've established our data ingestion strategy, where raw data from various enterprise sources (both SAP and non-SAP) is successfully ingested into SAP HANA Cloud via SAP Integration Suite.

Now, our primary focus is on the feature engineering phase for this data, which resides entirely within SAP HANA Cloud. We need to perform various transformations, including:

Calculating derived metrics (e.g., DaysSinceLastActivity, InteractionCount).

Aggregating historical data.

Leveraging capabilities for automated feature engineering.

Preparing the engineered features in a structured, consumable format for machine learning model training and inference.

Ensuring scalability, performance, governance, and reusability of these features.

We are specifically evaluating the best approach for this feature engineering. We see two primary paths within the SAP BTP landscape for data already in HANA Cloud:

Leveraging SAP HANA Cloud's In-built Machine Learning Capabilities: Utilizing its SQL/SQLScript, Predictive Analysis Library (PAL), Automated Predictive Library (APL), AutoML features, and Python Machine Learning Client directly within HANA Cloud.

Utilizing SAP Datasphere: Building and managing the feature engineering pipelines and feature store within Datasphere, which leverages HANA Cloud as its underlying engine.

Based on your experience with similar predictive projects on SAP BTP, we'd greatly appreciate your insights on the following:

For feature engineering on data already residing in SAP HANA Cloud, what are the key advantages and disadvantages of:

Relying primarily on SAP HANA Cloud's native ML and data manipulation capabilities?

Integrating SAP Datasphere into the architecture for feature engineering?

In which scenarios would you strongly recommend one approach over the other for a project like ours?

Are there specific types of feature engineering tasks (e.g., complex aggregations, specific derived metrics, automated feature generation) where one service truly excels when the data source is HANA Cloud?

Any architectural patterns, best practices, or practical experiences (including considerations for MLOps, governance, and long-term maintainability) would be highly valuable.

Thank you in advance for your contributions!

#SAPHANACloud, #SAPDatasphere, #MachineLearning, #FeatureEngineering

I m SAP SD want to learn SAP BUSINESS AI

2025-07-19T07:58:00.658000+02:00

where do i get this wit respective My sap sd and what to install and from where and where do i get content

Personalized Recommendations LMS: Need more clarity

2025-07-24T16:29:41.218000+02:00

I want to know more about this feature.

Personalized Recommendations on the New Learning Home Page | SAP Help Portal

In this it says "It includes their profile information, growth portfolio attributes, and their system interactions and activities."

What it means by Profile information and System Interactions? Will it cover all the modules to suggest courses?

and can you explain the interactions? what acts as an interaction in the system? also what information it takes from the Profile?

Do we have a detailed explanation for this feature?

Also, if my interest is Ex. Leadership, will it show all the courses related to Leadership? or just some courses.?

Also, Can we know why a particular course was recommended? is it because I searched for it, or completed a course or any other thing? Can a user know that?

SAP SuccessFactors Learning Machine Learning SAP Learning Solution

Build Error in GenAI Mail Insights: Develop a CAP application using GenAI...

2025-09-19T10:39:28.715000+02:00

GenAI Mail Insights 빌드 오류와 관련하여 SAP 티켓에 대한 간결하고 전문적인 영어 메시지는 다음과 같습니다.

제목: GenAI Mail Insights Mission의 빌드 오류: AzureOpenAiChatClient와 TypeScript 유형 비호환성

설명:
안녕하세요,

"GenAI 메일 인사이트: GenAI와 검색 증강 생성(RAG)을 활용한 CAP 애플리케이션 개발" 과제를 진행하던 중, GitHub 저장소를 복제하고 npm build를 실행해 보았습니다 . 하지만 여러 TypeScript 오류로 인해 빌드가 실패했습니다.

주요 오류 메시지는 다음과 같습니다.

텍스트

오류: srv/mail-insights-service.ts(385,52): 오류 TS2345: 'AzureOpenAiChatClient' 유형의 인수는 'BaseLanguageModelInterface<any, BaseLanguageModelCallOptions>' 유형의 매개변수에 할당할 수 없습니다. 'generatePrompt' 속성의 유형이 호환되지 않습니다. ... 'import("node_modules/@langchain/core/dist/language_models/base").BaseLanguageModelCallOptions' 유형은 'import("node_modules/@sap-ai-sdk/langchain/node_modules/@langchain/core/dist/language_models/base").BaseLanguageModelCallOptions' 유형에 할당할 수 없습니다. 'callbacks' 속성의 유형이 호환되지 않습니다. ...

이 문제는 @langchain/core가 여러 번 설치되어 TypeScript 유형이 충돌하기 때문에 발생하는 것으로 보입니다 . 하나는 @langchain/core에 직접 설치되고 다른 하나는 @Pa_Vi-ai-sdk/langchain 아래에 중첩되어 설치됩니다. 이로 인해 AzureOpenAiChatClient 와 예상 인터페이스 간에 호환되지 않는 유형 정의가 발생하여 빌드가 실패합니다.

Node.js 버전이 요구 사항대로 20 이상인지 확인했고, 저장소 종속성을 수정하지 않았습니다. 그럼에도 불구하고 오류가 지속되어 더 이상 작업을 진행할 수 없습니다.

이런 유형의 호환성 문제에 대한 도움을 주실 수 있나요? 아니면 이 문제를 해결하는 데 권장하는 해결책이나 종속성의 업데이트된 버전을 알려주실 수 있나요?

귀하의 지원에 감사드립니다.

원하시면 더욱 맞춤화하는 데 도움을 드릴 수 있습니다.