class ImageFeederTree:
'''
This class is used to feed the images of faces to the classifier,
and getting back the predictions.
This should be used after the classifier has been trained.
'''
def __init__(self):
self.tree_classifier = tree.DecisionTreeClassifier()
self.tree_obj = TreeClassifier()
self.predicted_results = []
self.array_of_images = []
def convertRawDataToTestData(self, list_of_images):
'''
This module is used to convert the list of images,
into a format that can be accepted by the classifier.
This should be called before getPrediction().
#Parameters: 'list_of_images' is the List of images for which predictions are needed.
#Return: None
'''
length_of_list = len(list_of_images)
for i in range(length_of_list):
img = list_of_images[i]
img = np.asarray(img)
img = img.flatten()
self.array_of_images.append(img)
self.array_of_images = np.asarray(self.array_of_images)
def getPrediction(self, pickle_name):
'''
This module loads the classifier and,
predicts the label of the images passed to it.
#Parameters: 'pickle_name' is the pickle file to be loaded
#Return: Array of predictions
'''
self.tree_classifier = self.tree_obj.loadClassifier(pickle_name)
self.predicted_results = self.tree_classifier.predict(
self.array_of_images)
return self.predicted_results
def testKNNClassifier(trainingSet_data, trainingSet_target, testSet_data, testSet_target):
classChoice = int(input("1. Knn Class\n 2. Decision Tree \n >"))
if classChoice == 2:
myClassifyer = TreeClassifier()
else:
myClassifyer = Classifier()
myClassifyer.train(trainingSet_data, trainingSet_target)
# predictions of classifier
predictionSet = myClassifyer.predict(testSet_data) # check
classifier = KNeighborsClassifier(n_neighbors=3)
classifier.fit(trainingSet_data, trainingSet_target)
predictions = classifier.predict(testSet_data)
# calculate percent correct
numCorrect = 0
numGivenCorrect = 0
for x in range(0, len(testSet_target)):
if predictionSet[x] == testSet_target[x]:
numCorrect += 1
if predictions[x] == testSet_target[x]:
numGivenCorrect += 1
percentCorrect = numCorrect / float(len(predictionSet))
percentGivenCorrect = numGivenCorrect / float(len(predictions))
print ("Using my kNN percent correct is :")
print("%.2f " %percentCorrect)
print ("or ", numCorrect, " /", len(predictionSet))
# for given classifier
print ("Given kNN percent correct is :")
print ("%.2f " %percentGivenCorrect)
print ("or ", numGivenCorrect, " /", len(predictions))
return
def __init__(self):
self.tree_classifier = tree.DecisionTreeClassifier()
self.tree_obj = TreeClassifier()
self.predicted_results = []
self.array_of_images = []
def runTreeTest():
data = []
targets = []
labels = []
choice = int(input("Choose Set: \n 1. Votes \n 2. Lenses \n 3. Iris \n> "))
if (choice == 3):
iris = datasets.load_iris()
data = iris.data
targets = iris.target
data = convertToNom(data)
labels = ["SepalL", "SepalW", "PetalL", "PetalW"]
elif (choice == 2):
lensesData = readLensData("lenses.csv")
np_data = np.array(lensesData)
data = np_data[:, :(len(lensesData[0]) - 1)]
targets = np_data[:, (len(lensesData[0]) - 1)] # last column
labels = ["Age", "Precription", "Astigmatic", "TearProduction"]
else:
votesData = readVotesData("votes.csv")
np_data = np.array(votesData)
data = np_data[:, -(len(votesData[0]) - 1):] #5 digits from the left
targets = np_data[:, 0] # first column republican1 or democrate 0
labels = ["handicapped-infants", "water-project-cost-sharing", "adoption-of-the-budget-resolution",
"physician-fee-freeze", "el-salvador-aid", "religious-groups-in-schools", "anti-satellite-test-ban",
"aid-to-nicaraguan-contras","mx-missile","immigration","synfuels-corporation-cutback","education-spending",
"superfund-right-to-sue","crime","duty-free-exports","export-administration-act-south-africa"]
# randomize
np.unique(targets)
np.random.seed(0)
# 3 randomize the index
indices = np.random.permutation(len(data))
testNum = int(len(data) * 0.3) # 70% train, 30% test
trainingSet_data = data[indices[:-int(testNum)]]
trainingSet_target = targets[indices[:-int(testNum)]]
testSet_data = data[indices[-int(testNum):]]
testSet_target = targets[indices[-int(testNum):]]
train_classes = []
test_classes = []
# convert to arrays
for t in trainingSet_target:
train_classes.append(t)
for t in testSet_target:
test_classes.append(t)
#print("Train Data:\n", trainingSet_data, train_classes)
#print("Test Data:\n", testSet_data, test_classes)
classifier = TreeClassifier()
train_labels = []
train_labels = list(labels)
classifier.train(trainingSet_data, train_classes, labels)
myPredictions = classifier.predict(testSet_data, train_labels)
# calculate percent correct
numCorrect = 0
numGivenCorrect = 0
for x in range(0, len(testSet_target)):
if myPredictions[x] == test_classes[x]:
numCorrect += 1
#if predictionSet[x] == testSet_target[x]:
# numGivenCorrect += 1
percentCorrect = numCorrect / float(len(test_classes))
#percentGivenCorrect = numGivenCorrect / float(len(predictions))
print ("Using my decision tree :")
print("%.2f " %percentCorrect)
print ("or ", numCorrect, " /", len(test_classes))
return
