Package PyML :: Package classifiers :: Module platt :: Class Platt
[frames] | no frames]

Class Platt

source code

base.pymlObject.PyMLobject --+        
                             |        
    baseClassifiers.Classifier --+    
                                 |    
     composite.CompositeClassifier --+
                                     |
                                    Platt

Converts a real valued classifier into a conditional probability estimator. This is achieved by fitting a sigmoid with parameters A and B to the values of the decision function: f(x) --> 1/(1+exp(A*f(x)+B)

code is a based on Platt's pseudocode from:

John C. Platt. Probabilistic Outputs for Support Vector Machines and Comparisons to Regularized Likelihood Methods. in: Advances in Large Margin Classifiers A. J. Smola, B. Schoelkopf, D. Schuurmans, eds. MIT Press (1999).

:Keywords:

Nested Classes
    Inherited from baseClassifiers.Classifier
  resultsObject
Instance Methods
 
train(self, data, **args) source code
 
fit_A_B(self, prior1, prior0, out, deci, Y) source code
 
decisionFunc(self, data, i) source code
 
classify(self, data, i) source code
 
test(classifier, data, **args)
test a classifier on a given dataset
source code
 
save(self, fileName) source code
 
load(self, fileName) source code
    Inherited from composite.CompositeClassifier
 
__init__(self, classifier, **args) source code
 
__repr__(self) source code
 
getTest(self) source code
 
preproject(self, data) source code
 
setTest(self) source code
    Inherited from baseClassifiers.Classifier
 
cv(classifier, data, numFolds=5, **args)
perform k-fold cross validation
source code
 
getTrainingTime(self) source code
 
logger(self) source code
 
loo(classifier, data, **args)
perform Leave One Out
source code
 
nCV(classifier, data, **args)
runs CV n times, returning a 'ResultsList' object.
source code
 
project(self, data)
project a test dataset to the training data features.
source code
 
stratifiedCV(classifier, data, numFolds=5, **args)
perform k-fold stratified cross-validation; in each fold the number of patterns from each class is proportional to the relative fraction of the class in the dataset
source code
 
trainFinalize(self) source code
 
trainTest(classifierTemplate, data, trainingPatterns, testingPatterns, **args)
Train a classifier on the list of training patterns, and test it on the test patterns
source code
 
twoClassClassify(self, data, i) source code
Class Variables
  attributes = {'mode': 'holdOut', 'numFolds': 3, 'fittingFracti...
    Inherited from composite.CompositeClassifier
  deepcopy = True
    Inherited from baseClassifiers.Classifier
  type = 'classifier'
Method Details

train(self, data, **args)

source code 
Overrides: baseClassifiers.Classifier.train

decisionFunc(self, data, i)

source code 
Overrides: composite.CompositeClassifier.decisionFunc

classify(self, data, i)

source code 
Overrides: baseClassifiers.Classifier.classify

test(classifier, data, **args)

source code 
test a classifier on a given dataset
Parameters:
  • classifier - a trained classifier
  • data - a dataset
  • stats - whether to compute the statistics of the match between the predicted labels and the given labels [True by default]
Returns:
a Results class instance
Overrides: evaluators.assess.test

save(self, fileName)

source code 
Overrides: baseClassifiers.Classifier.save

Class Variable Details

attributes

Value:
{'mode': 'holdOut', 'numFolds': 3, 'fittingFraction': 0.2}