Quick Start¶
30-Second Start¶
That's it. The pipeline will:
- Load your data
- Run EDA
- Clean the data
- Engineer features
- Compare 14+ models
- Tune the best model
- Save a production-ready pipeline
What You Get Back¶
# Best model name
print(ctx.best_model_name) # "lightgbm"
# All metrics
print(ctx.metrics)
# {'test_r2': 0.847, 'test_mae': 0.292, 'test_rmse': 0.447, ...}
# Saved model path
print(ctx.reports["model_path"]) # "artifacts/model_v1.joblib"
Load and Use the Saved Model¶
import joblib
import pandas as pd
# Load the saved model
model = joblib.load("artifacts/model_v1.joblib")
# Prepare new data
new_data = pd.DataFrame({
"feature1": [1.0, 2.0, 3.0],
"feature2": ["a", "b", "c"],
})
# Predict
predictions = model.predict(new_data)
CLI Alternative¶
Next Steps¶
- Python API — Full API documentation
- Configuration — Customize the pipeline
- YAML Config — Config-driven pipelines