def test_submission(self):
train_data = extract_features(load_adult_train_data())
valid_data = extract_features(load_adult_valid_data())
model = submission(train_data)
predictions = [predict(model, p) for p in train_data]
print
print
print "Training Accuracy:", accuracy(train_data, predictions)
predictions = [predict(model, p) for p in valid_data]
print "Validation Accuracy:", accuracy(valid_data, predictions)
print
def test_submission(self):
train_data = extract_features(load_adult_train_data())
valid_data = extract_features(load_adult_valid_data())
model = submission(train_data)
predictions = [predict(model, p) for p in train_data]
print()
print()
print("Training Accuracy:", accuracy(train_data, predictions))
predictions = [predict(model, p) for p in valid_data]
print("Validation Accuracy:", accuracy(valid_data, predictions))
print()
def test_submission(self):
"""Overall test.
"""
train_data = extract_features(load_adult_train_data())
valid_data = extract_features(load_adult_valid_data())
model = submission(train_data)
predictions = [predict(model, p) for p in train_data]
print("Training Accuracy: {0}".format(
accuracy(train_data, predictions)))
predictions = [predict(model, p) for p in valid_data]
print("Validation Accuracy: {0}".format(
accuracy(valid_data, predictions)))
def test_submission():
valid_data = test_data
model = submission(train_data)
predictions = [predict(model, p) for p in train_data]
print
print
# print predictions
train_accuracy = accuracy(train_data, predictions)
print "Training Accuracy:", train_accuracy
# print train_data
predictions = [predict(model, p) for p in valid_data]
valid_accuracy = accuracy(valid_data, predictions)
print "Validation Accuracy:", valid_accuracy
print
return train_accuracy, valid_accuracy
def test_neural_net(self):
print "Testing neural net.."
print
#getting data using 80% of data as training.
#If you want to test soething other than 3 vs 5, just change the input to loadmnist.
#For example, loadmnist(1, 7)
data = loadmnist(3, 5)
train_data = data[:int(len(data)*0.8)]
validation_data =data[int(len(data)*0.8):]
#train the model
m = neural_net(train_data)
#evaluate
predictions = [m.predict(p) for p in train_data]
print "Training Accuracy:", accuracy(train_data, predictions)
predictions = [m.predict(p) for p in validation_data]
print "Validation Accuracy:", accuracy(validation_data, predictions)
def test_neural_net(self):
print "Testing neural net.."
print
#getting data using 80% of data as training.
#If you want to test soething other than 3 vs 5, just change the input to loadmnist.
#For example, loadmnist(1, 7)
data = loadmnist(3, 5)
train_data = data[:int(len(data)*0.8)]
validation_data =data[int(len(data)*0.8):]
#train the model
m = neural_net(train_data)
#evaluate
predictions = [m.predict(p) for p in train_data]
print "Training Accuracy:", accuracy(train_data, predictions)
predictions = [m.predict(p) for p in validation_data]
print "Validation Accuracy:", accuracy(validation_data, predictions)
def test_accuracy(self):
data = extract_features(load_adult_train_data())
a = accuracy(data, [0.4]*len(data))
self.assertAlmostEqual(a, 0.751077514754)
def test_accuracy(self):
data = extract_features(load_adult_train_data())
a = accuracy(data, [0]*len(data))
self.assertAlmostEqual(a, 0.751077514754)
def test_accuracy(self):
data = [dict([('label',np.matrix([1]))]) for i in range(25)]+[dict([('label',np.matrix([0]))]) for i in range(75)]
a = accuracy(data, [np.matrix([0]) for i in range(len(data))])
self.assertAlmostEqual(a, 0.75)
max_num_faces = max(max_num_faces, len(item_dict))
# predict for result of one picture
predictions = []
for i in range(0, len(item_dict)):
face = {}
face["faceRectangle"] = item_dict[i]["faceRectangle"]
face["result"] = (predict(
model, extract_features_single_point(item_dict[i]["scores"]))
>= 0.5)
predictions.append(face)
predictions_all.append(predictions)
print max_num_faces
return predictions_all, max_num_faces
if __name__ == "__main__":
# prepare the model
train_data = extract_features(load_adult_train_data())
model = submission(train_data)
print model
predictions = [predict(model, p) for p in train_data]
print
print
print "Training Accuracy:", accuracy(train_data, predictions)
app.run(host='ec2-52-206-17-234.compute-1.amazonaws.com',
port=8000,
threaded=True)
def test_accuracy():
data = train_data
a = accuracy(data, [0] * len(data))
def test_accuracy(self):
"""Tests the accuracy calculation.
"""
data = extract_features(load_adult_train_data())
a = accuracy(data, [0] * len(data))
self.assertAlmostEqual(a, 0.7636129)
def test_accuracy(self):
data = [dict([('label',np.matrix([1]))]) for i in range(25)]+[dict([('label',np.matrix([0]))]) for i in range(75)]
a = accuracy(data, [np.matrix([0]) for i in range(len(data))])
self.assertAlmostEqual(a, 0.75)
