main.py

#!/usr/bin/env python

import numpy as np
import random

from triager.classifier import models, kernels, selectors, tests, utils
from triager.classifier.parsers import Label
from triager import parsers


def main():
    # Parse data into documents
    #parser = parsers.MRSParser("data/prop/MRs/", project_match="OPW.*")
    parser = parsers.CSVBugzillaParser("data/opensource/netbeans")
    #parser = parsers.BugzillaParser("data/opensource/mozilla_firefox")
    print "Parsing data by parser: %s" % parser
    documents = parser.parse()
    # Shuffle documents
    random.seed(4) # use 3 for MRS
    random.shuffle(documents)
    random.seed()
    # Filter unlabeled documents and documents that are not labeled with
    # a class that occurs at lest `min_class_occur`
    documents = [doc for doc in documents if doc.label]
    print "Filtering unlabeled documents..."
    min_class_occur = 30
    documents = utils.filter_docs(documents, min_class_occur=min_class_occur)
    print "Filtering documents labeled by class that occurs less than " \
        + "%s times." % min_class_occur
    # Split between train and cv data
    n = len(documents)
    split_pct = 7/10.0
    split_x = int(np.ceil(n*split_pct))
    docs_train = documents[:split_x]
    docs_cv = documents[split_x:]
    # Create model
    selector = selectors.TFIDFDecorator(selectors.StopWordsDecorator(
        selectors.BasicSelector()))
    kernel = kernels.GaussianKernel()
    model = models.SVMModel(feature_selector=selector, kernel=kernel, C=240)
    print "Created model: %s." % model
    # Train model
    print "Training model on %s instances..." % len(docs_train)
    model.train(docs_train)
    print "Number of classes is: %s" % len(model.feature_selector.labels)

    # Test model (accuracy)
    print "Computing accuracy for train set (size=%s)..." % len(docs_train)
    accuracy_train = tests.accuracy(model, docs_train)
    print "Computing accuracy for CV set (size=%s)..." % len(docs_cv)
    accuracy_cv = tests.accuracy(model, docs_cv)
    print "Accuracy of train set is: '%.4f'." % accuracy_train
    print "Accuracy of CV set is: '%.4f'." % accuracy_cv
    # Test model (precision and recall)
    print "Computing macro-average precision and recall for train set..."
    pr_train = tests.precision_and_recall(model, docs_train)
    print "Computing macro-average precision and recall for CV set..."
    pr_cv = tests.precision_and_recall(model, docs_cv)
    print "Average precision and recall of train set is: '%.4f' and '%.4f'." \
        % pr_train
    print "Average precision and recall of CV set is: '%.4f' and '%.4f'." \
        % pr_cv

if __name__ == "__main__":
    try:
        main()
    except KeyboardInterrupt as e:
        pass
    finally:
        print "----------"