Skip to main content
Help us improve PyPI by participating in user testing. All experience levels needed!

BREW: Python Multiple Classifier System API

Project description

https://badge.fury.io/py/brew.png https://travis-ci.org/viisar/brew.png?branch=master Code Health https://coveralls.io/repos/github/viisar/brew/badge.svg?branch=master Join the chat at https://gitter.im/viisar/brew

brew: A Multiple Classifier Systems API

This project was started in 2014 by Dayvid Victor and Thyago Porpino
for the Multiple Classifier Systems class at Federal University of Pernambuco.
The aim of this project is to provide an easy API for Ensembling, Stacking,
Blending, Ensemble Generation, Ensemble Pruning, Dynamic Classifier Selection,
and Dynamic Ensemble Selection.

Features

  • General: Ensembling, Stacking and Blending.
  • Ensemble Classifier Generators: Bagging, Random Subspace, SMOTE-Bagging, ICS-Bagging, SMOTE-ICS-Bagging.
  • Dynamic Selection: Overall Local Accuracy (OLA), Local Class Accuracy (LCA), Multiple Classifier Behavior (MCB), K-Nearest Oracles Eliminate (KNORA-E), K-Nearest Oracles Union (KNORA-U), A Priori Dynamic Selection, A Posteriori Dynamic Selection, Dynamic Selection KNN (DSKNN).
  • Ensemble Combination Rules: majority vote, min, max, mean and median.
  • Ensemble Diversity Metrics: Entropy Measure E, Kohavi Wolpert Variance, Q Statistics, Correlation Coefficient p, Disagreement Measure, Agreement Measure, Double Fault Measure.
  • Ensemble Pruning: Ensemble Pruning via Individual Contribution (EPIC).

Example

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
import itertools

import sklearn

from sklearn.linear_model import LogisticRegression
from sklearn.svm import SVC
from sklearn.ensemble import RandomForestClassifier

from brew.base import Ensemble, EnsembleClassifier
from brew.stacking.stacker import EnsembleStack, EnsembleStackClassifier
from brew.combination.combiner import Combiner

from mlxtend.data import iris_data
from mlxtend.evaluate import plot_decision_regions

# Initializing Classifiers
clf1 = LogisticRegression(random_state=0)
clf2 = RandomForestClassifier(random_state=0)
clf3 = SVC(random_state=0, probability=True)

# Creating Ensemble
ensemble = Ensemble([clf1, clf2, clf3])
eclf = EnsembleClassifier(ensemble=ensemble, combiner=Combiner('mean'))

# Creating Stacking
layer_1 = Ensemble([clf1, clf2, clf3])
layer_2 = Ensemble([sklearn.clone(clf1)])

stack = EnsembleStack(cv=3)

stack.add_layer(layer_1)
stack.add_layer(layer_2)

sclf = EnsembleStackClassifier(stack)

clf_list = [clf1, clf2, clf3, eclf, sclf]
lbl_list = ['Logistic Regression', 'Random Forest', 'RBF kernel SVM', 'Ensemble', 'Stacking']

# Loading some example data
X, y = iris_data()
X = X[:,[0, 2]]

# Plotting Decision Regions
gs = gridspec.GridSpec(2, 3)
fig = plt.figure(figsize=(10, 8))

itt = itertools.product([0, 1, 2], repeat=2)

for clf, lab, grd in zip(clf_list, lbl_list, itt):
    clf.fit(X, y)
    ax = plt.subplot(gs[grd[0], grd[1]])
    fig = plot_decision_regions(X=X, y=y, clf=clf, legend=2)
    plt.title(lab)
plt.show()
decision regions plots

Dependencies

  • Python 2.7+
  • scikit-learn >= 0.15.2
  • Numpy >= 1.6.1
  • SciPy >= 0.9
  • Matplotlib >= 0.99.1 (examples, only)
  • mlxtend (examples, only)

Installing

You can easily install brew using pip:

pip install brew

or, if you prefer an up-to-date version, get it from here:

pip install git+https://github.com/viisar/brew.git

Important References

  • Kuncheva, Ludmila I. Combining pattern classifiers: methods and algorithms. John Wiley & Sons, 2014.
  • Zhou, Zhi-Hua. Ensemble methods: foundations and algorithms. CRC Press, 2012.

Documentation

The full documentation is at http://brew.rtfd.org.

History

0.1.0 (2014-11-12)

  • First release on PyPI.

Project details


Release history Release notifications

This version
History Node

0.1.4

History Node

0.1.3

History Node

0.1.2

History Node

0.1.1

History Node

0.1.0

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Filename, size & hash SHA256 hash help File type Python version Upload date
brew-0.1.4.zip (48.8 kB) Copy SHA256 hash SHA256 Source None Oct 11, 2016

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging CloudAMQP CloudAMQP RabbitMQ AWS AWS Cloud computing Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page