def test_euclidean_distance_1dimension(self):
"""Tests the euclidean distance in one dimension"""
dist_from_origin = lambda v: euclidean_distance([0], v)
self.assertEqual(100, dist_from_origin([100]))
self.assertEqual(100, dist_from_origin([-100]))
self.assertEqual(123, dist_from_origin([123]))
self.assertEqual(123, dist_from_origin([-123]))
self.assertEqual(0, dist_from_origin([0]))
self.assertEqual(0, dist_from_origin([0]))
self.assertEqual(1, dist_from_origin([-1]))
self.assertEqual(1, dist_from_origin([1]))
dist_from_initial = lambda v: euclidean_distance([77], v)
self.assertEqual(33, dist_from_initial([77 + 33]))
self.assertEqual(33, dist_from_initial([77 - 33]))
self.assertEqual(11, dist_from_initial([77 + 11]))
self.assertEqual(11, dist_from_initial([77 - 11]))
self.assertEqual(123, dist_from_initial([77 + 123]))
self.assertEqual(123, dist_from_initial([77 - 123]))
def test_euclidean_distance_2dimension(self):
"""Tests the euclidean distance in 2 dimensions"""
dist_from_origin = lambda v: euclidean_distance([0, 0], v)
# Do it first with known whole numbers
self.assertEqual(5, dist_from_origin([3, 4]))
self.assertEqual(5, dist_from_origin([-3, -4]))
self.assertEqual(13, dist_from_origin([5, 12]))
self.assertEqual(13, dist_from_origin([-5, -12]))
self.assertEqual(15, dist_from_origin([9, 12]))
self.assertEqual(15, dist_from_origin([-9, -12]))
self.assertEqual(52, dist_from_origin([20, 48]))
self.assertEqual(52, dist_from_origin([-20, -48]))
# Next we will do it with a couple doubles, but we need
# to round
self.assertEqual(8.6023, round(dist_from_origin([5, 7]), 4))
self.assertEqual(8.6023, round(dist_from_origin([-5, -7]), 4))
self.assertEqual(13.0384, round(dist_from_origin([7, 11]), 4))
self.assertEqual(13.0384, round(dist_from_origin([-7, -11]), 4))
self.assertEqual(17.0294, round(dist_from_origin([11, 13]), 4))
self.assertEqual(17.0294, round(dist_from_origin([-11, -13]), 4))
dist_from_initial = lambda v: euclidean_distance([7, 23], v)
# Do it first with known whole numbers
self.assertEqual(5, dist_from_initial([7 + 3, 23 + 4]))
self.assertEqual(5, dist_from_initial([7 - 3, 23 - 4]))
self.assertEqual(13, dist_from_initial([7 + 5, 23 + 12]))
self.assertEqual(13, dist_from_initial([7 - 5, 23 - 12]))
self.assertEqual(15, dist_from_initial([7 + 9, 23 + 12]))
self.assertEqual(15, dist_from_initial([7 - 9, 23 - 12]))
self.assertEqual(52, dist_from_initial([7 + 20, 23 + 48]))
self.assertEqual(52, dist_from_initial([7 + 20, 23 + 48]))
# Next we do it with some doubles which require rounding
self.assertEqual(8.6023, round(dist_from_initial([7 + 5, 23 + 7]), 4))
self.assertEqual(8.6023, round(dist_from_initial([7 - 5, 23 - 7]), 4))
self.assertEqual(13.0384, round(dist_from_initial([7 + 7, 23 + 11]), 4))
self.assertEqual(13.0384, round(dist_from_initial([7 - 7, 23 - 11]), 4))
self.assertEqual(17.0294, round(dist_from_initial([7 + 11, 23 + 13]), 4))
self.assertEqual(17.0294, round(dist_from_initial([7 - 11, 23 - 13]), 4))
def get_guess(self, vector):
# If we haven't obtained atleast k values of
# our window then we should use true instead
if len(self._training_data) < self._k:
return True
# grab the k nearest neighbors using euclidean
# distance
dist = lambda x: euclidean_distance(vector, x[1])
closest_points = nsmallest(self._k, enumerate(self._training_data), key=dist)
closest_labels = [self._training_labels[i] for (i, pt) in closest_points]
return max(set(closest_labels), key=closest_labels.count)
def test_euclidean_distance_Ndimension(self):
"""Tests the euclidean distance in 2+ dimensions"""
self.assertEqual(15, euclidean_distance([0, 0, 0], [10, 10, 5]))
self.assertEqual(15, euclidean_distance([0, 0, 0], [-10, -10, -5]))
self.assertEqual(17, euclidean_distance([0, 0, 0, 0], [10, 10, 8, 5]))
self.assertEqual(17, euclidean_distance([0, 0, 0, 0], [-10, -10, -8, -5]))
self.assertEqual(8, euclidean_distance([0, 0, 0, 0, 0], [5, 1, 1, 1, 6]))
self.assertEqual(8, euclidean_distance([0, 0, 0, 0, 0], [-5, -1, -1, -1, -6]))
def classify(vector):
dist = lambda x: euclidean_distance(vector, x[1])
closest_points = nsmallest(self._k, enumerate(data), key=dist)
closest_labels = [labels[i] for (i, pt) in closest_points]
return max(set(closest_labels), key=closest_labels.count)
def classify(vector):
dist = lambda x: euclidean_distance(vector, x[1])
closest_points = nsmallest(self._k, enumerate(data), key=dist)
closest_labels = [labels[i] for (i, pt) in closest_points]
return max(set(closest_labels), key=closest_labels.count)
def index_of_nearest_centroid(self, vector):
dist = lambda v: euclidean_distance(v[1], vector)
return min(enumerate(self._centroids), key=dist)[0]
