def write_results(data, dist, small, train, LOOCV):
filename = dist
if small:
filename += '_small'
filename += '_train' if train else '_test'
if LOOCV:
filename += '_LOOCV'
np.savetxt(f'./results/QE/{filename}.txt', data, delimiter=",", fmt='%s')
train = np.loadtxt(r'./data/MNIST_train.csv', delimiter=',')
X_train, y_train = train[:, 1:], train[:, 0]
cls = kNN(X=X_train, y=y_train)
while True:
if info.size == 0:
last_index = int(0)
else:
last_index = int(info[-1][0])
if last_index == 21_000:
print('DONE!')
break
print('last_index', last_index, '/', len(X_train))
y_hat = cls.predict(X_train[last_index + 1:last_index + 101, :],
'minkowski',
LOOCV=True,
p=p)
def classify_euklides(self, test_value: StructNeighbour, k: int):
knn = kNN(kNN.distance_euklides)
return knn.assign(self.neighbourhood, test_value, k)
def classify_street(self, test_value: StructNeighbour, k: int):
knn = kNN(kNN.distance_street)
return knn.assign(self.neighbourhood, test_value, k)
def a_classify_street(self, test_value: StructNeighbour, kmin: int,
kmax: int):
knn = kNN(kNN.distance_street)
return knn.assign_array(self.neighbourhood, test_value, kmin, kmax)
def classify_chebyshev(self, test_value: StructNeighbour, k: int):
knn = kNN(kNN.distance_chebyshev)
return knn.assign(self.neighbourhood, test_value, k)
