scikit-learn Template

The scikit-learn template scaffolds a production-ready ML project with scikit-learn, pandas, and traditional ML algorithms. Two variants are available:

sklearn: basic model template
sklearn-pipeline: full preprocessing + model pipeline

Quick Start

cirron init my-sklearn-model --template sklearn

The model type (classification, regression, etc.) is selected interactively during cirron init.

Project Structure

my-sklearn-model/
├── src/
│   ├── model.py           # create_model() + preprocess_data()
│   ├── data_loader.py     # CSV loading + train/test split
│   ├── train.py           # Training script
│   └── inference.py       # Inference / prediction
├── data/
├── models/                # joblib-serialized model + preprocessor
├── requirements.txt
├── Dockerfile
└── cirron.yaml

For shared files (Dockerfile, .gitignore, README.md, .env.example, tests), see Common files.

Cirron-specific bits

cirron.yaml

framework: sklearn
model_type: classification
train_entrypoint: src/train.py
inference_entrypoint: src/inference.py

Profiling hooks

The generated train.py and inference.py integrate with the Cirron SDK. Call ci.profile() once at module top, then wrap the estimator with ci.wrap() to record fit, predict, and pipeline-step spans:

import cirron as ci
from sklearn.ensemble import RandomForestClassifier

ci.profile()

self.model = ci.wrap(RandomForestClassifier(n_estimators=100))
self.model.fit(self.X_train_processed, self.y_train)

requirements.txt

scikit-learn>=1.3.0
numpy>=1.21.0
pandas>=1.5.0
matplotlib>=3.5.0
joblib>=1.2.0

Model Types

Classification: defaults to RandomForestClassifier(n_estimators=100, max_depth=10). Metrics: accuracy, precision, recall, F1.
Regression: defaults to RandomForestRegressor(n_estimators=100, max_depth=10). Metrics: MSE, R².

Swap the default model in src/model.py:

def create_model():
    return SVC(probability=True, random_state=42)

Model Persistence

The generated train.py saves both the model and the fitted preprocessor together via joblib:

joblib.dump({
    'model': self.model,
    'preprocessor': self.preprocessor,
    'config': self.config,
}, 'models/model.joblib')

inference.py reloads both so prediction-time preprocessing matches training.

Usage

python src/train.py     # Train
cirron build            # Container
cirron deploy           # Deploy

Hyperparameter Tuning

Wrap create_model() with GridSearchCV for sweep-based tuning:

from sklearn.model_selection import GridSearchCV

def create_model():
    base = RandomForestClassifier(random_state=42)
    return GridSearchCV(base, {
        'n_estimators': [50, 100, 200],
        'max_depth': [5, 10, None],
    }, cv=5)

Next Steps

PyTorch Template

Deep learning framework

TensorFlow Template

Alternative deep learning framework

Common files

Dockerfile, tests, and shared scaffolding

Deployment

Deploy your trained model

​scikit-learn Template

​Quick Start

​Project Structure

​Cirron-specific bits

​cirron.yaml

​Profiling hooks

​requirements.txt

​Model Types

​Model Persistence

​Usage

​Hyperparameter Tuning

​Next Steps