def classifyNB(vec2Classify, n_name):
trainMatrix, trainCategory, allIds = getModelByGroup(
n_name, "dispersed", trainModel, originModel["ids"], groupNum,
testPercentage)
model = tree.DecisionTreeClassifier()
model.fit(trainMatrix, trainCategory)
predicted = model.predict(vec2Classify)
return predicted[0]
def classifyNB(vec2Classify, n_name):
dating_data_mat_linear, testCategory, allIds = getModelByGroup(
n_name, "dispersed", trainModel, originModel["ids"], groupNum,
testPercentage)
norm_mat_linear = z_score_norm(dating_data_mat_linear)
knn_classifier = KNeighborsClassifier()
knn_classifier.fit(norm_mat_linear, testCategory)
predicted = knn_classifier.predict(vec2Classify)
return predicted[0]
def dating_class_test(n_name):
print("\ngroup name is " + n_name)
testMatrix, testCategory, allIds = getModelByGroup(n_name, "test",
trainModel,
originModel["ids"],
groupNum,
testPercentage)
right_count = 0.0
for key, line in enumerate(testMatrix):
cls = classifyNB([line], n_name)
actual = testCategory[key]
notice = ""
if (cls == actual):
right_count += 1.0
if (cls != actual):
notice = "fail"
cha = cls - actual
too_many = ""
pid = 0
if len(allIds) > 0:
pid = allIds[key]
if cha >= errorMoney or cha <= -errorMoney:
too_many = "many"
print(
"pid is %s, the classifier came back with: %s, the real answer is: %s %s %s"
% (pid, cls, actual, notice, too_many))
#print key,line,cls
rate = right_count / len(testCategory)
print("the total right rate is: %f" % (rate))
print("error num is: %f" % int((len(testCategory) - right_count)))
print("right num is: %f" % int(right_count))
return rate