SAP HANA PAL Hands-On — Classification, Clustering and Association with UnifiedClassification
As of 2026-09-25
Hands-on module on SAP HANA's Predictive Analysis Library (PAL), the in-database, non-generative machine-learning engine behind classic SAP predictive scenarios. Covers what PAL is and why it runs inside HANA (SAP's own trillion-row rationale), its ten data-mining categories, the two ways to call it (SQLScript procedures and the hana-ml Python client), classification with UnifiedClassification (HybridGradientBoostingTree, RandomDecisionTree), clustering (KMeans, density-based methods) and association mining (Apriori), preprocessing discipline, PAL's own AutoML and massive-training classes, model evaluation and comparison, and the boundaries with APL (M339), SAP-RPT and the generative AI hub. Three exercises (classify, cluster and mine associations, choose PAL/APL/SAP-RPT) and a self-assessment gate the move to M339.
What you will learn
- Explain what PAL is, why it runs inside SAP HANA rather than on a client, and name its ten data-mining categories with one SAP use case per category
- Connect to SAP HANA Cloud with hana-ml's ConnectionContext and build a HANA DataFrame on a labelled SAP-shaped table without moving the data to the client
- Fit, score and interpret a UnifiedClassification model (HybridGradientBoostingTree or RandomDecisionTree), including its feature-importance output
- Run KMeans clustering and Apriori association mining on SAP-shaped data and state the business question each result actually answers
- Profile and clean input columns before fitting, and use PAL's model evaluation and comparison capability to choose between algorithms
- Choose between PAL, APL and SAP-RPT for a given business question, and describe how a trained PAL model reaches production through ISLM
Module overview
Who this is for. You have completed M333 (AI & LLM Fundamentals for SAP Consultants) or already know the vocabulary of the generative-AI path — tokens, embeddings, grounding — and you now need the other half of "SAP AI": prediction on tables, computed where the data already lives. This module is hands-on with the Predictive Analysis Library (PAL), the classic, non-generative machine-learning engine built into SAP HANA. M339 does the same for its automated sibling, the Automated Predictive Library (APL); M340 goes deeper into the Python client, hana-ml, that wraps both; M341 shows how S/4HANA industrialises a trained model through Intelligent Scenario Lifecycle Management (ISLM). Work through this module with a HANA Cloud instance (trial or client sandbox) if you can reach one; every exercise also has a paper-only path.
Prerequisites
- M333 (AI & LLM Fundamentals for SAP Consultants) or an equivalent working vocabulary of tokens, embeddings and grounding
- Comfort reading SQL and basic Python; prior exposure to pandas or scikit-learn is useful but not required
- Optional for the hands-on variants of E1-E3: access to a SAP HANA Cloud trial or a client sandbox with PAL enabled, and the hana-ml package installed
Outcomes
- Explain to a client why an in-database algorithm avoids a costly data-movement step, using PAL's own trillion-row rationale.
- Fit and evaluate a PAL classification model on labelled SAP data end to end, from connection to feature importance.
- Run and interpret a clustering exercise and an association-mining exercise on SAP-shaped data, stating the business question each answers.
- Recommend PAL, APL or SAP-RPT for a given SAP prediction use case, with the trade-off written down rather than assumed.
Full module available to members. The full module adds: the decision framework · the end-to-end scenario walkthrough · the KPI scorecard · the anti-patterns · the code blocks · the knowledge check · the diagrams.