def dbscan(eps, minpts, point_list):
counter = MeowCounter()
cluster_list = []
for p in point_list:
if p.label is None:
p.label = "visited"
counter.count()
n_eps = find_n(p, eps, point_list)
if len(n_eps) < minpts:
p.label = "noise"
else:
cluster = Cluster()
cluster_list.append(cluster)
cluster.add_point(p)
while len(n_eps) > 0:
p_p = n_eps.pop()
if p_p.label is None:
p_p.label = "visited"
counter.count()
n_eps_2 = find_n(p_p, eps, point_list)
if len(n_eps_2) >= minpts:
for i in n_eps_2:
if i.label is None:
n_eps.append(i)
if p_p.cluster is None:
cluster.add_point(p_p)
return cluster_list
def dbscan(self, data):
self.init_params()
self.data = data
## Setting up the plot
fig = plt.figure()
axis_proj = 'rectilinear'
if self.dim > 2:
axis_proj = '%dd' % self.dim
ax = fig.add_subplot(111, projection = axis_proj)
#default noise cluster
noise = Cluster('Noise', self.dim)
self.clusters.add(noise)
for point in data:
if point not in self.visited:
self.visited.append(point)
neighbour_pts = self.region_query(point)
if len(neighbour_pts) < self.min_pts:
noise.add_point(point)
else:
name = 'cluster-%d' % self.cluster_count
new_cluster = Cluster(name, self.dim)
self.cluster_count += 1
self.expand_cluster(new_cluster, point, neighbour_pts)
if self.dim == 2:
ax.scatter(new_cluster.get_X(), new_cluster.get_Y(), c = self.color[self.cluster_count % len(self.color)],
marker = 'o', label = name)
elif self.dim == 3:
ax.scatter(new_cluster.get_X(), new_cluster.get_Y(), new_cluster.get_Z(), marker = 'o',
c = self.color[self.cluster_count % len(self.color)], label = name)
ax.hold(True)
if len(noise.get_points()) != 0:
if self.dim > 2:
ax.scatter(noise.get_X(), noise.get_Y(), noise.get_Z(), marker = 'x', label = noise.name)
else:
ax.scatter(noise.get_X(), noise.get_Y(), marker = 'x', label = noise.name)
print ("Number of clusters found: %d" % self.cluster_count)
ax.hold(False)
ax.legend(loc='lower left')
ax.grid(True)
plt.title(r'DBSCAN Clustering', fontsize=18)
plt.show()
def dbscan(self, data):
self.init_params()
self.data = data
## Setting up the plot
# fig = plt.figure()
axis_proj = 'rectilinear'
if self.dim > 2:
axis_proj = '%dd' % self.dim
# ax = fig.add_subplot(111, projection = axis_proj)
# default noise cluster
noise = Cluster('Noise', self.dim)
self.clusters.add(noise)
for point in data:
if point not in self.visited:
self.visited.append(point)
neighbour_pts = self.region_query(point)
if len(neighbour_pts) < self.min_pts:
noise.add_point(point)
else:
name = 'cluster-%d' % self.cluster_count
new_cluster = Cluster(name, self.dim)
self.cluster_count += 1
self.expand_cluster(new_cluster, point, neighbour_pts)
# if self.dim == 2:
# ax.scatter(new_cluster.get_X(), new_cluster.get_Y(), c = self.color[self.cluster_count % len(self.color)],
# marker = 'o', label = name)
# elif self.dim == 3:
# ax.scatter(new_cluster.get_X(), new_cluster.get_Y(), new_cluster.get_Z(), marker = 'o',
# c = self.color[self.cluster_count % len(self.color)], label = name)
# if len(noise.get_points()) != 0:
# if self.dim > 2:
# ax.scatter(noise.get_X(), noise.get_Y(), noise.get_Z(), marker = 'x', label = noise.name)
# else:
# ax.scatter(noise.get_X(), noise.get_Y(), marker = 'x', label = noise.name)
print("Number of clusters found: %d" % self.cluster_count)
def dbscan(self, data):
self.init_params()
self.data = data
## Setting up the plot
fig = plt.figure()
axis_proj = 'rectilinear'
if self.dim > 2:
axis_proj = '%dd' % self.dim
ax = fig.add_subplot(111, projection=axis_proj)
#default noise cluster
noise = Cluster('Noise', self.dim)
self.clusters.add(noise)
for point in data:
if point not in self.visited:
self.visited.append(point)
neighbour_pts = self.region_query(point)
if len(neighbour_pts) < self.min_pts:
noise.add_point(point)
else:
name = 'cluster-%d' % self.cluster_count
new_cluster = Cluster(name, self.dim)
self.cluster_count += 1
self.expand_cluster(new_cluster, point, neighbour_pts)
if self.dim == 2:
ax.scatter(new_cluster.get_X(),
new_cluster.get_Y(),
c=self.color[self.cluster_count %
len(self.color)],
marker='o',
label=name)
elif self.dim == 3:
ax.scatter(new_cluster.get_X(),
new_cluster.get_Y(),
new_cluster.get_Z(),
marker='o',
c=self.color[self.cluster_count %
len(self.color)],
label=name)
ax.hold(True)
if len(noise.get_points()) != 0:
if self.dim > 2:
ax.scatter(noise.get_X(),
noise.get_Y(),
noise.get_Z(),
marker='x',
label=noise.name)
else:
ax.scatter(noise.get_X(),
noise.get_Y(),
marker='x',
label=noise.name)
print("Number of clusters found: %d" % self.cluster_count)
ax.hold(False)
ax.legend(loc='lower left')
ax.grid(True)
plt.title(r'DBSCAN Clustering', fontsize=18)
plt.show()
