def test(self, labels, test_set):
_,ts = helper.format_for_scikit(labels, test_set)
predictions = self.classifier.predict(ts)
if self.plot_roc:
print("ROC curve plot unavailable for %s") % (str(self))
return helper.accuracy(labels, predictions), predictions
def train_helper(self, labels, train_set, kernel_name, deg=3):
self.classifier = svm.SVC(kernel=kernel_name, degree=deg,
probability=True, cache_size=1000.0)
l,ts = helper.format_for_scikit(labels, train_set)
#pca = PCA(n_components='mle')
#ts = pca.fit_transform(ts)
self.classifier.fit(ts, l)
def test(self, labels, test_set):
l,ts = helper.format_for_scikit(labels, test_set)
#pca = PCA(n_components='mle')
#ts = pca.fit_transform(ts)
predictions = self.classifier.predict(ts)
if self.plot_roc:
probas = self.classifier.predict_proba(ts)
helper.roc(probas, l, str(self))
return helper.accuracy(labels, predictions), predictions
def test(self, labels, test_set):
_,ts = helper.format_for_scikit(labels, test_set)
predictions = self.classifier.predict(ts)
if self.plot_roc:
feat_list = test_set[0].keys()
# FIXME: handle output file name
outfile = '../data/dt.dot'
print("ROC curve unavailable for this classifier.\n" +
"Creating a Decision Tree plot instead in: %s") % (outfile)
tree.export_graphviz(self.classifier, outfile, feat_list)
return helper.accuracy(labels, predictions), predictions
def chi2(data, labels, print_results=True):
"""Compute chi-squared statistic for each class/feature combination.
The results are printed (if the `print_results` paramter is True) in
the stdout AND returned."""
y, X = helper.format_for_scikit(labels, data)
chi2, pval = fs.chi2(X,y)
feature_names = [name for name in data[0].keys()]
if print_results:
print('Chi-squared values for each feature :')
for i,v in enumerate(chi2):
print('%s => %f') % (feature_names[i], v)
print('#####################################')
print('p-value for each feature :')
for i,v in enumerate(pval):
print('%s => %f') % (feature_names[i], v)
return chi2, pval
def train(self, labels, train_set):
self.classifier = lda.LDA()
l,ts = helper.format_for_scikit(labels, train_set)
self.classifier.fit(ts, l)
def train(self, labels, train_set):
self.classifier = tree.DecisionTreeClassifier()
l,ts = helper.format_for_scikit(labels, train_set)
self.classifier.fit(ts, l)
def train(self, labels, train_set):
self.classifier = BernoulliNB()
l, ts = helper.format_for_scikit(labels, train_set)
self.classifier.fit(ts, l)
def train(self, labels, train_set):
self.classifier = linear_model.LogisticRegression()
l,ts = helper.format_for_scikit(labels, train_set)
self.classifier.fit(ts, l)
