AgentSkillsCN

sap-hana-ml

SAP HANA 机器学习 Python 客户端（hana-ml）开发技能。 适用于：使用 SAP HANA 内置的机器学习功能，借助 Python hana-ml 库中的 PAL/APL 算法、DataFrame 操作、AutoML、模型持久化与可视化，构建 ML 解决方案。 关键词：hana-ml、SAP HANA、机器学习、PAL、APL、预测分析、HANA DataFrame、ConnectionContext、分类、回归、聚类、时间序列、ARIMA、梯度提升、AutoML、SHAP、模型存储。

SKILL.md
--- frontmatter
name: sap-hana-ml
description: |
  SAP HANA Machine Learning Python Client (hana-ml) development skill.
  
  Use when: Building ML solutions with SAP HANA's in-database machine learning
  using Python hana-ml library for PAL/APL algorithms, DataFrame operations,
  AutoML, model persistence, and visualization.
  
  Keywords: hana-ml, SAP HANA, machine learning, PAL, APL, predictive analytics,
  HANA DataFrame, ConnectionContext, classification, regression, clustering,
  time series, ARIMA, gradient boosting, AutoML, SHAP, model storage
license: GPL-3.0
metadata:
  version: 1.1.0
  last_verified: 2025-11-27
  package_version: 2.22.241011

SAP HANA ML Python Client (hana-ml)

Package Version: 2.22.241011
Last Verified: 2025-11-27

Table of Contents

Installation & Setup

bash
pip install hana-ml

Requirements: Python 3.8+, SAP HANA 2.0 SPS03+ or SAP HANA Cloud

Quick Start

Connection & DataFrame

python
from hana_ml import ConnectionContext

# Connect
conn = ConnectionContext(
    address='<hostname>',
    port=443,
    user='<username>',
    password='<password>',
    encrypt=True
)

# Create DataFrame
df = conn.table('MY_TABLE', schema='MY_SCHEMA')
print(f"Shape: {df.shape}")
df.head(10).collect()

PAL Classification

python
from hana_ml.algorithms.pal.unified_classification import UnifiedClassification

# Train model
clf = UnifiedClassification(func='RandomDecisionTree')
clf.fit(train_df, features=['F1', 'F2', 'F3'], label='TARGET')

# Predict & evaluate
predictions = clf.predict(test_df, features=['F1', 'F2', 'F3'])
score = clf.score(test_df, features=['F1', 'F2', 'F3'], label='TARGET')

APL AutoML

python
from hana_ml.algorithms.apl.classification import AutoClassifier

# Automated classification
auto_clf = AutoClassifier()
auto_clf.fit(train_df, label='TARGET')
predictions = auto_clf.predict(test_df)

Model Persistence

python
from hana_ml.model_storage import ModelStorage

ms = ModelStorage(conn)
clf.name = 'MY_CLASSIFIER'
ms.save_model(model=clf, if_exists='replace')

Core Libraries

PAL (Predictive Analysis Library)

  • •100+ algorithms executed in-database
  • •Categories: Classification, Regression, Clustering, Time Series, Preprocessing
  • •Key classes: UnifiedClassification, UnifiedRegression, KMeans, ARIMA
  • •See: references/PAL_ALGORITHMS.md for complete list

APL (Automated Predictive Library)

  • •AutoML capabilities with automatic feature engineering
  • •Key classes: AutoClassifier, AutoRegressor, GradientBoostingClassifier
  • •See: references/APL_ALGORITHMS.md for details

DataFrames

  • •Lazy evaluation - builds SQL until collect() called
  • •In-database processing for optimal performance
  • •See: references/DATAFRAME_REFERENCE.md for complete API

Visualizers

  • •EDA plots, model explanations, metrics
  • •SHAP integration for model interpretability
  • •See: references/VISUALIZERS.md for 14 visualization modules

Common Patterns

Train-Test Split

python
from hana_ml.algorithms.pal.partition import train_test_val_split

train, test, val = train_test_val_split(
    data=df,
    training_percentage=0.7,
    testing_percentage=0.2,
    validation_percentage=0.1
)

Feature Importance

python
# APL models
importance = auto_clf.get_feature_importances()

# PAL models
from hana_ml.algorithms.pal.preprocessing import FeatureSelection
fs = FeatureSelection()
fs.fit(train_df, features=features, label='TARGET')

Pipeline

python
from hana_ml.algorithms.pal.pipeline import Pipeline
from hana_ml.algorithms.pal.preprocessing import Imputer, FeatureNormalizer

pipeline = Pipeline([
    ('imputer', Imputer(strategy='mean')),
    ('normalizer', FeatureNormalizer()),
    ('classifier', UnifiedClassification(func='RandomDecisionTree'))
])

Best Practices

  1. •Use lazy evaluation - Operations build SQL without execution until collect()
  2. •Leverage in-database processing - Keep data in HANA for performance
  3. •Use Unified interfaces - Consistent APIs across algorithms
  4. •Save models - Use ModelStorage for persistence
  5. •Explain predictions - Use SHAP explainers for interpretability
  6. •Monitor AutoML - Use PipelineProgressStatusMonitor for long-running jobs

Bundled Resources

Reference Files

  • •

    references/DATAFRAME_REFERENCE.md (479 lines)

    • •ConnectionContext API, DataFrame operations, SQL generation
  • •

    references/PAL_ALGORITHMS.md (869 lines)

    • •Complete PAL algorithm reference (100+ algorithms)
    • •Classification, Regression, Clustering, Time Series, Preprocessing
  • •

    references/APL_ALGORITHMS.md (534 lines)

    • •AutoML capabilities, automated feature engineering
    • •AutoClassifier, AutoRegressor, GradientBoosting classes
  • •

    references/VISUALIZERS.md (704 lines)

    • •14 visualization modules (EDA, SHAP, metrics, time series)
    • •Plot types, configuration, export options
  • •

    references/SUPPORTING_MODULES.md (626 lines)

    • •Model storage, spatial analytics, graph algorithms
    • •Text mining, statistics, error handling

Error Handling

python
from hana_ml.ml_exceptions import Error

try:
    clf.fit(train_df, features=features, label='TARGET')
except Error as e:
    print(f"HANA ML Error: {e}")

Documentation