def predict(self, features):
"""
predict the label given the feature
:param features: list[list]
:return: labels
"""
parentList = self.__BN.getParents()
featureList = [featureAttri.name for featureAttri in self._featureTable]
predY = []
postProbY = []
for i in range(len(features)):
postProbY.append(copy.deepcopy(self.__P_y))
for j in range(len(features[i])):
featureIndex = self._featureTable[j].values.index(features[i][j])
if j == 0:
for yIndex in range(len(self._featureTable[-1].values)):
postProbY[i][yIndex] *= self.__P_x_p12[j][yIndex][featureIndex]
else:
p1 = featureList.index(parentList[self._featureTable[j].name][0])
p1Index = self._featureTable[p1].values.index(features[i][p1])
for yIndex in range(len(self._featureTable[-1].values)):
postProbY[i][yIndex] *= self.__P_x_p12[j][p1Index][yIndex][featureIndex]
utils.uniformize(postProbY[i])
predY.append(self._featureTable[-1].values[postProbY[i].index(max(postProbY[i]))])
return predY, postProbY
def predict(self, features):
"""
predict the label given the feature
:param features: list[list]
:return: labels, postProbY
"""
labels = []
postProbY = []
for inst in features:
postProbY.append(copy.deepcopy(self.__prob_y))
for i in range(len(self._featureTable[-1].values)):
for j in range(len(inst)):
postProbY[-1][i] *= self.__prob_x_y[i][j][inst[j]]
utils.uniformize(postProbY[-1])
labels.append(self._featureTable[-1].values[postProbY[-1].index(max(postProbY[-1]))])
return labels, postProbY
