otter-streams-xgboost

Gradient Boosting

High-performance gradient boosting inference for trained models from tabular datasets

Module Overview

The XGBoost module provides native XGBoost inference optimized for tabular and structured data problems. It supports multiple model formats, batch predictions, and GPU acceleration for high-throughput gradient boosting inference.

🔷

XGBoost Engine

Gradient Boosting

Native XGBoost inference optimized for tabular and structured data problems.

  • Binary, JSON, and UBJSON model formats
  • Regression, binary, and multi-class classification
  • Efficient batch prediction via DMatrix
  • GPU acceleration when available
⚡

Performance Features

Optimization

Advanced performance optimizations for high-throughput inference.

  • Thread-safe Booster predictions
  • Column-major DMatrix for efficiency
  • Native missing value handling (NaN)
  • Automatic memory management

Implementing XGBoost Inference

Guide to deploying XGBoost models for real-time predictions.

  1. Add Maven Dependency 
    <dependency>
    <groupId>com.codedstreams</groupId>
    <artifactId>otter-streams-xgboost</artifactId>
    <version>1.0.16</version>
</dependency>
  2. Train and Save XGBoost Model 
    import xgboost as xgb

# Train model
dtrain = xgb.DMatrix(X_train, label=y_train)
params = {'objective': 'binary:logistic', 'max_depth': 3}
model = xgb.train(params, dtrain, num_boost_round=100)

# Save in binary format (fastest)
model.save_model('model.bin')

# Or save in JSON format (portable)
model.save_model('model.json')
  3. Configure and Use Engine 
    ModelConfig config = ModelConfig.builder()
    .modelPath("model.xgb")
    .modelId("xgboost-classifier")
    .format(ModelFormat.XGBOOST_BINARY)
    .build();

XGBoostInferenceEngine engine = new XGBoostInferenceEngine();
engine.initialize(config);

// Single prediction
Map<String, Object> inputs = Map.of(
    "age", 35.0f,
    "income", 75000.0f,
    "credit_score", 720.0f
);

InferenceResult result = engine.infer(inputs);
float prediction = (float) result.getOutput("prediction");
  4. Batch Predictions 
    // Batch inference for efficiency
Map<String, Object>[] batchInputs = new Map[1000];
// ... populate batch inputs

InferenceResult batchResult = engine.inferBatch(batchInputs);
float[][] predictions = (float[][]) batchResult.getOutput("batch_predictions");

Quick Links

🔷 XGBoost Engine Details ⚡ Performance Features 📚 Full JavaDocs 🏠 Back to Home 🔷 ← PyTorch Module 🔷 PMML Module →

Maven Dependency

<dependency>
    <groupId>com.codedstreams</groupId>
    <artifactId>otter-streams-xgboost</artifactId>
    <version>1.0.16</version>
</dependency>