Stripje - Make sklearn pipelines lean

Speed up your scikit-learn pipelines for single-row predictions by 2-10x!

Stripje is a high-performance compiler that converts trained scikit-learn pipelines into optimized Python functions, eliminating numpy overhead for single-row inference.

🚀 Why Stripje?

• ⚡ 2-200x faster single-row predictions, depending on the pipeline complexity
• 🔧 Drop-in replacement - works with your existing pipelines
• 🎯 Zero configuration - just compile and use
• 🛠️ Production ready - optimized for real-time inference

📦 Installation

pip install stripje

Or with uv (recommended):

uv add stripje

⚡ Quick Start

from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
from stripje import compile_pipeline

# 1. Create and fit your pipeline as usual
pipeline = Pipeline([
    ('scaler', StandardScaler()),
    ('classifier', LogisticRegression())
])
pipeline.fit(X_train, y_train)

# 2. Compile for fast single-row inference
fast_predict = compile_pipeline(pipeline)

# 3. Get predictions up to 10x faster!
test_row = [1.2, -0.5, 0.8, -1.1]
prediction = fast_predict(test_row)  # Much faster than pipeline.predict([test_row])

🎯 The Problem We Solve

Standard scikit-learn pipelines are slow for single predictions because they're optimized for batch processing. When you need to predict one row at a time (like in web APIs), numpy operations create unnecessary overhead.

Stripje compiles your trained pipeline into a specialized function that:

• ✅ Extracts fitted parameters once
• ✅ Eliminates array creation overhead
• ✅ Uses native Python operations
• ✅ Maintains identical results

📊 Performance Comparison

import time
from sklearn.datasets import make_classification
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression

# Setup
X, y = make_classification(n_samples=1000, n_features=20)
pipeline = Pipeline([('scaler', StandardScaler()), ('clf', LogisticRegression())])
pipeline.fit(X, y)
fast_predict = compile_pipeline(pipeline)

test_row = X[0].tolist()

# Benchmark single-row predictions
def benchmark_standard():
    start = time.time()
    for _ in range(1000):
        pipeline.predict([test_row])
    return time.time() - start

def benchmark_compiled():
    start = time.time()
    for _ in range(1000):
        fast_predict(test_row)
    return time.time() - start

standard_time = benchmark_standard()
compiled_time = benchmark_compiled()
speedup = standard_time / compiled_time

print(f"Standard pipeline: {standard_time:.3f}s")
print(f"Compiled pipeline: {compiled_time:.3f}s")
print(f"Speedup: {speedup:.1f}x faster!")

🔧 Supported Components

Stripje supports the most commonly used scikit-learn components:

🔄 Transformers

• Scalers: StandardScaler, MinMaxScaler, RobustScaler, MaxAbsScaler
• Encoders: OneHotEncoder, OrdinalEncoder, LabelEncoder
• Other: Normalizer, QuantileTransformer, SelectKBest

🎯 Estimators

• Classification: LogisticRegression, RandomForestClassifier, DecisionTreeClassifier, GaussianNB
• Regression: LinearRegression

🏗️ Composite

• ColumnTransformer - Full support with nested compilation

More components coming soon! See Contributing to request or add support.

📖 More Examples

Complex Pipeline with ColumnTransformer

from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.ensemble import RandomForestClassifier

# Create a complex pipeline
preprocessor = ColumnTransformer([
    ('num', StandardScaler(), ['age', 'income']),
    ('cat', OneHotEncoder(), ['category', 'region'])
])

pipeline = Pipeline([
    ('preprocessor', preprocessor),
    ('classifier', RandomForestClassifier(n_estimators=10))
])

# Fit and compile
pipeline.fit(X_train, y_train)
fast_predict = compile_pipeline(pipeline)

# Single-row prediction
row = [25, 50000, 'A', 'North']  # [age, income, category, region]
prediction = fast_predict(row)

Real-World API Usage

from flask import Flask, request, jsonify
import joblib

app = Flask(__name__)

# Load and compile your model once at startup
model = joblib.load('trained_pipeline.pkl')
fast_predict = compile_pipeline(model)

@app.route('/predict', methods=['POST'])
def predict():
    data = request.json['features']
    prediction = fast_predict(data)  # Super fast!
    return jsonify({'prediction': prediction.tolist()})

🚫 Limitations

• Input must be lists/arrays (no pandas DataFrames directly)
• No sparse matrix support
• Some transformers use approximations (e.g., QuantileTransformer)
• Only listed components are supported

📚 API Reference

compile_pipeline(pipeline)

Compiles a fitted scikit-learn pipeline into a fast single-row prediction function.

Args:

• pipeline: A fitted scikit-learn Pipeline

Returns:

• Function that takes a single row (list/array) and returns predictions

Raises:

• ValueError: If pipeline contains unsupported components

get_supported_transformers()

Returns list of all supported transformer/estimator classes.

📁 Examples & Benchmarks

Check out the examples/ directory for:

• simple_example.py - Basic usage
• benchmark.py - Performance comparisons
• comprehensive_benchmark.py - Detailed benchmarks
• profiler_demo.py - Profiling tools

🔌 Extending Support

Want to add support for a new transformer? It's easy:

from stripje import register_step_handler

@register_step_handler(YourTransformer)
def handle_your_transformer(step):
    # Extract parameters from the fitted step
    param1 = step.param1_
    param2 = step.param2_

    def transform_one(x):
        # Implement single-row transformation logic
        result = []
        for val in x:
            # Your transformation logic here
            transformed_val = val * param1 + param2
            result.append(transformed_val)
        return result

    return transform_one

🤝 Contributing

Contributions are welcome! Please feel free to submit a pull request or open an issue.

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

🛠️ Development

Setup Development Environment

1. Clone the repository:

git clone https://github.com/hadi-gharibi/stripje.git
cd stripje

2. Install all dependencies (including optional ones for full testing):

uv sync --all-extras

3. Install pre-commit hooks:

uv run pre-commit install

Code Quality Tools

This project uses modern Python development tools:

• Ruff: Fast linting, formatting, and import sorting
• MyPy: Static type checking
• pre-commit: Automated code quality checks

Run code quality checks:

# Lint and auto-fix issues
uv run ruff check src/ tests/ --fix

# Format code
uv run ruff format src/ tests/

# Type checking
uv run mypy src/

# Run all pre-commit hooks
uv run pre-commit run --all-files

Testing

Run tests:

uv run pytest

Run tests with coverage:

uv run pytest --cov=stripje

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.