本文整理匯總了Python中KNNClassifierRegion.KNNClassifierRegion.getLatestDistances方法的典型用法代碼示例。如果您正苦於以下問題：Python KNNClassifierRegion.getLatestDistances方法的具體用法？Python KNNClassifierRegion.getLatestDistances怎麽用？Python KNNClassifierRegion.getLatestDistances使用的例子？那麽, 這裏精選的方法代碼示例或許可以為您提供幫助。您也可以進一步了解該方法所在類KNNClassifierRegion.KNNClassifierRegion的用法示例。
在下文中一共展示了KNNClassifierRegion.getLatestDistances方法的1個代碼示例，這些例子默認根據受歡迎程度排序。您可以為喜歡或者感覺有用的代碼點讚，您的評價將有助於係統推薦出更棒的Python代碼示例。
示例1: KNNAnomalyClassifierRegion
# 需要導入模塊: from KNNClassifierRegion import KNNClassifierRegion [as 別名]
# 或者: from KNNClassifierRegion.KNNClassifierRegion import getLatestDistances [as 別名]
#.........這裏部分代碼省略.........
self._knnclassifier._knn.removeIds(idsToDelete.tolist())
assert self._knnclassifier._knn._numPatterns == nProtos - len(idsToDelete)
def _recomputeRecordFromKNN(self, record):
"""
returns the classified labeling of record
"""
inputs = {
"categoryIn": [None],
"bottomUpIn": self._getStateAnomalyVector(record),
}
outputs = {"categoriesOut": numpy.zeros((1,)),
"bestPrototypeIndices":numpy.zeros((1,)),
"categoryProbabilitiesOut":numpy.zeros((1,))}
# Only use points before record to classify and after the wait period.
classifier_indexes = numpy.array(
self._knnclassifier.getParameter('categoryRecencyList'))
valid_idx = numpy.where(
(classifier_indexes >= self.getParameter('trainRecords')) &
(classifier_indexes < record.ROWID)
)[0].tolist()
if len(valid_idx) == 0:
return None
self._knnclassifier.setParameter('inferenceMode', None, True)
self._knnclassifier.setParameter('learningMode', None, False)
self._knnclassifier.compute(inputs, outputs)
self._knnclassifier.setParameter('learningMode', None, True)
classifier_distances = self._knnclassifier.getLatestDistances()
valid_distances = classifier_distances[valid_idx]
if valid_distances.min() <= self._classificationMaxDist:
classifier_indexes_prev = classifier_indexes[valid_idx]
rowID = classifier_indexes_prev[valid_distances.argmin()]
indexID = numpy.where(classifier_indexes == rowID)[0][0]
category = self._knnclassifier.getCategoryList()[indexID]
return category
return None
def _labelToCategoryNumber(self, label):
"""
Since the KNN Classifier stores categories as numbers, we must store each
label as a number. This method converts from a label to a unique number.
Each label is assigned a unique bit so multiple labels may be assigned to
a single record.
"""
if label not in self.saved_categories:
self.saved_categories.append(label)
return pow(2, self.saved_categories.index(label))
def _labelListToCategoryNumber(self, labelList):
"""
This method takes a list of labels and returns a unique category number.
This enables this class to store a list of categories for each point since
the KNN classifier only stores a single number category for each record.
"""
categoryNumber = 0
for label in labelList:
categoryNumber += self._labelToCategoryNumber(label)
return categoryNumber