https://github.com/bettercallshao/sklearn_surrogatesearchcv

Surrogate adaptive randomized search for hyper-parameters tuning in sklearn.
https://github.com/bettercallshao/sklearn_surrogatesearchcv

cross-validation data-science hyper-parameter-tuning machine-learning optimization pysot python sklearn surrogate-based-optimization

Last synced: 3 days ago
JSON representation

Surrogate adaptive randomized search for hyper-parameters tuning in sklearn.

• Host: GitHub
• URL: https://github.com/bettercallshao/sklearn_surrogatesearchcv
• Owner: bettercallshao
• License: mit
• Created: 2018-12-10T01:19:50.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2019-10-13T06:55:38.000Z (almost 5 years ago)
• Last Synced: 2024-09-13T23:53:39.883Z (20 days ago)
• Topics: cross-validation, data-science, hyper-parameter-tuning, machine-learning, optimization, pysot, python, sklearn, surrogate-based-optimization
• Language: Python
• Homepage:
• Size: 20.5 KB
• Stars: 2
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Surrogate Search CV

[![CircleCI](https://circleci.com/gh/bettercallshao/sklearn_surrogatesearchcv.svg?style=shield)](https://circleci.com/gh/bettercallshao/sklearn_surrogatesearchcv)

[![PyPi](https://badge.fury.io/py/sklearn-surrogatesearchcv.svg)](https://badge.fury.io/py/sklearn-surrogatesearchcv)

This package implements a randomized hyper parameter search for sklearn (similar to `RandomizedSearchCV`) but utilizes surrogate adaptive sampling from pySOT. Use this similarly to GridSearchCV with a few extra paramters.

## Usage

```

pip install sklearn-surrogatesearchcv

```

The interface is unimaginative, stylistically similar to `RandomizedSearchCV`.

```

class SurrogateSearchCV(object):

    """Surrogate search with cross validation for hyper parameter tuning.

    """

    def __init__(self, estimator, n_iter=10, param_def=None, refit=False,

                 **kwargs):

        """

        :param estimator: estimator

        :param n_iter: number of iterations to run (default 10)

        :param param_def: list of dictionaries, e.g.

            [

                {

                    'name': 'alpha',

                    'integer': False,

                    'lb': 0.1,

                    'ub': 0.9,

                },

                {

                    'name': 'max_depth',

                    'integer': True,

                    'lb': 3,

                    'ub': 12,

                }

            ]

        :param **: every other parameter is the same as GridSearchCV

        """

```

The result can be found in the following properties of the class instance after running.

```

params_history_

score_history_

best_params_

best_score_

```

For a complete example, please refer to `src/test/test_basic.py`.

## Resources

A slide about role of surrogate optimization in ml. [link](https://www.slideshare.net/TimTan2/machine-learning-vs-traditional-optimization)