def _get_min_sq_dist(self, point: Vector) -> float:
min_sq_dist = math.inf
for centroid in self._centroids:
if centroid != []:
sq_dist = euclidean_distance(point, centroid)**2
if sq_dist < min_sq_dist:
min_sq_dist = sq_dist
return min_sq_dist
def _get_dist(self, c1: Cluster, c2: Cluster, measure=min) -> float:
"""
Get the distance between to cluster c1 and c2 with euclidean distance.
Using the min function as default measure(single link).
"""
distances = []
for i in c1:
for j in c2:
dist = euclidean_distance(self._dataset[i], self._dataset[j])
distances.append(dist)
return measure(distances)
def predict(self, point: Vector) -> int:
min_dist = math.inf
best_label = -1
for i in range(self._k):
cluster_vectors = [self._dataset[j] for j in self._clusters[i]]
dist = sum(euclidean_distance(point, p) for p in cluster_vectors)
if dist < min_dist:
min_dist = dist
best_label = i
print(f"The label of {point} is {best_label}")
return best_label
def predict(self, point: Vector) -> int:
"""
Predict the label of the new sample point
"""
min_dist = math.inf
label = -1
for cluster_label, cluster in enumerate(self._clusters):
dist = min(
[euclidean_distance(point, self._dataset[i]) for i in cluster])
if dist < min_dist:
min_dist = dist
label = cluster_label
return label
def _get_centroid_label(self, point: Vector,
centroids: DataSet) -> int:
distances = [euclidean_distance(point, centroid)
for centroid in centroids]
return distances.index(min(distances))