Milk is a machine learning toolkit in Python.

Its focus is on supervised classification with several classifiers available: SVMs (based on libsvm), k-NN, random forests, decision trees. It also performs feature selection. These classifiers can be combined in many ways to form different classification systems.

For unsupervised learning, milk supports k-means clustering and affinity propagation.

Milk is flexible about its inputs. It optimised for numpy arrays, but can often handle anything (for example, for SVMs, you can use any dataype and any kernel and it does the right thing).

There is a strong emphasis on speed and low memory usage. Therefore, most of the performance sensitive code is in C++. This is behind Python-based interfaces for convenience.

To learn more, check the docs at http://packages.python.org/milk/ or the code demos included with the source at milk/demos/.

Here is how to test how well you can classify some features,labels data, measured by cross-validation:

import numpy as np
import milk
features = np.random.rand(100,10) # 2d array of features: 100 examples of 10 features each
labels = np.zeros(100)
features[50:] += .5
labels[50:] = 1
confusion_matrix, names = milk.nfoldcrossvalidation(features, labels)
print 'Accuracy:', confusion_matrix.trace()/float(confusion_matrix.sum())

If want to use a classifier, you instanciate a learner object and call its train() method:

import numpy as np
import milk
features = np.random.rand(100,10)
labels = np.zeros(100)
features[50:] += .5
labels[50:] = 1
learner = milk.defaultclassifier()
model = learner.train(features, labels)

# Now you can use the model on new examples:
example = np.random.rand(10)
print model.apply(example)
example2 = np.random.rand(10)
example2 += .5
print model.apply(example2)

There are several classification methods in the package, but they all use the same interface: train() returns a model object, which has an apply() method to execute on new instances.

License: MIT

Author: Luis Pedro Coelho (with code from LibSVM and scikits.learn)

API Documentation: http://packages.python.org/milk/

Mailing List: http://groups.google.com/group/milk-users

• SVMs. Using the libsvm solver with a pythonesque wrapper around it.
• LASSO
• K-means using as little memory as possible. It can cluster millions of instances efficiently.
• Random forests
• Self organising maps
• Stepwise Discriminant Analysis for feature selection.
• Non-negative matrix factorisation
• Affinity propagation

The ChangeLog file contains a more complete history.

• Add subspace projection kNN
• Export pdist in milk namespace
• Add Eigen to source distribution
• Add measures.curves.roc
• Add mds_dists function
• Add verbose argument to milk.tests.run

• Add coordinate-descent based LASSO
• Add unsupervised.center function
• Make zscore work with NaNs (by ignoring them)
• Propagate apply_many calls through transformers
• Much faster SVM classification with means a much faster defaultlearner() [measured 2.5x speedup on yeast dataset!]

• Add select_n_best & rank_corr to featureselection
• Add Euclidean MDS
• Add tree multi-class strategy
• Fix adaboost with boolean weak learners (issue #6, reported by audy (Austin Richardson))
• Add axis arguments to zscore()

• Make defaultlearner able to take extra arguments
• Make ctransforms_model a supervised_model (adds apply_many)
• Add expanded argument to defaultlearner
• Fix corner case in SDA
• Fix repeated_kmeans
• Fix parallel gridminimise on Windows
• Add multi_label argument to normaliselabels
• Add multi_label argument to nfoldcrossvalidation.foldgenerator
• Do not fork a process in gridminimise if nprocs == 1 (makes for easier debugging, at the cost of slightly more complex code).
• Add milk.supervised.multi_label
• Fix ext.jugparallel when features is a Task
• Add milk.measures.bayesian_significance

• Fix important bug in multi-process gridsearch

• Use multiprocessing to take advantage of multi core machines (off by default).
• Add perceptron learner
• Set random seed in random forest learner
• Add warning to milk/__init__.py if import fails
• Add return value to gridminimise
• Set random seed in precluster_learner
• Implemented Error-Correcting Output Codes for reduction of multi-class to binary (including probability estimation)
• Add multi_strategy argument to defaultlearner()
• Make the dot kernel in svm much, much, faster
• Make sigmoidal fitting for SVM probability estimates faster
• Fix bug in randomforest (patch by Wei on milk-users mailing list)

For older versions, see ChangeLog file