lazytune 0.1.2

Fast hyperparameter optimization using screening and pruning

LazyTune

LazyTune is a fast and efficient hyperparameter optimization framework for machine learning models.

It dramatically reduces unnecessary training time by using a smart screening → ranking → pruning → full training pipeline — while staying 100% compatible with scikit-learn estimators.

Supports classification & regression, all scikit-learn metrics, custom scorers, cross-validation screening, early pruning of poor configurations, parallel execution, and clean result reporting.

Features

• Compatible with any scikit-learn-style estimator
• Works for both classification and regression
• Supports all scikit-learn built-in metrics (accuracy, f1, r2, neg_mean_squared_error, etc.)
• Allows custom scoring functions via make_scorer
• Fast initial screening with cross-validation
• Early pruning of weak hyperparameter settings (prune_ratio)
• Parallel execution support (n_jobs)
• Structured trial summaries and ranking
• Returns best model, parameters, score + detailed report

Installation

pip install lazytune

Quick Start Example

from sklearn.datasets import load_breast_cancer
from sklearn.ensemble import RandomForestClassifier
from lazytune import SmartSearch

X, y = load_breast_cancer(return_X_y=True)

param_grid = {
    "n_estimators": [50, 100, 150, 200],
    "max_depth": [5, 10, 15, None],
    "min_samples_split": [2, 3, 4, 5]
}

search = SmartSearch(
    estimator=RandomForestClassifier(random_state=42),
    param_grid=param_grid,
    metric="accuracy",
    cv_folds=3,
    prune_ratio=0.5,       # keep top 50% after screening
    n_jobs=-1              # use all available cores
)

search.fit(X, y)

print("Best parameters:", search.best_params_)
print("Best CV score:   ", search.best_score_)
print("\nBest model:\n", search.best_estimator_)

More Examples

SVM Classification

from sklearn.svm import SVC
from lazytune import SmartSearch

search = SmartSearch(
    estimator=SVC(random_state=42),
    param_grid={
        "C": [0.1, 1, 10, 50, 100],
        "kernel": ["linear", "rbf"],
        "gamma": ["scale", "auto", 0.001, 0.0001]
    },
    metric="f1_macro",
    cv_folds=5,
    prune_ratio=0.6
)

Regression (Random Forest)

from sklearn.ensemble import RandomForestRegressor
from lazytune import SmartSearch

search = SmartSearch(
    estimator=RandomForestRegressor(random_state=42),
    param_grid={
        "n_estimators": [100, 200, 300, 500],
        "max_depth": [8, 12, 16, None],
        "min_samples_split": [2, 4, 8]
    },
    metric="r2",
    cv_folds=4,
    n_jobs=-1
)

Supported Metrics (examples)

Classification

accuracy • f1 • f1_macro • f1_weighted • precision • recall • roc_auc • balanced_accuracy • ...

Regression

r2 • neg_mean_squared_error • neg_root_mean_squared_error • neg_mean_absolute_error • neg_mean_absolute_percentage_error • ...

Custom metrics → use sklearn.metrics.make_scorer

How It Works (LazyTune Strategy)

1. Generate all (or sampled) hyperparameter combinations
2. Quick screening round with cross-validation (low resources)
3. Rank configurations by performance
4. Prune bottom performers (controlled by prune_ratio)
5. Train remaining promising candidates more thoroughly
6. Return best model + full summary of all evaluated trials

→ Much faster than full GridSearchCV while usually keeping very similar final performance.

Main API – SmartSearch

Key Attributes

Attribute	Description
best_params_	Best found hyperparameter dictionary
best_score_	Best cross-validated score
best_estimator_	Fully fitted estimator with best parameters
summary_	pandas DataFrame with trial results & rankings
cv_results_	Detailed cross-validation results per candidate

Main Methods

• .fit(X, y)
• .predict(X)
• .score(X, y)
• .get_params() / .set_params()

Requirements

• Python ≥ 3.8
• numpy
• pandas
• scikit-learn

Author

Anik Chand

License

MIT License

---

Feedback, issues, stars, and contributions are very welcome!
Happy tuning! 🚀