def cluster_set(data_obj, centers, iterations, name):
print(name)
data_obj.labels = [0 for i in buildings.coords]
clustering = deepcopy(data_obj)
num = len(centers)
print("A")
#clustering.c_positions = [ trivial_gen.gauss_point(i,0.001) for i in centers]
clustering.c_positions = centers
clustering.c_number = num
clustering = partitioning.k_means(clustering,
initialisation="default_pos",
iterations=iterations)
visualise(clustering, name)
print("B")
comparison = deepcopy(data_obj)
comparison.c_number = num
comparison = partitioning.k_means(comparison,
initialisation="kmeans_++",
iterations=iterations)
comparison.labes = [i + 1 for i in clustering.labels]
out = {}
out["number"] = len(centers)
out["comparison"] = comparative.scikit_all_scores_dict(
clustering, comparison)
out["default_pos"] = standalone.standard_scores_dict(clustering)
out["kmeans_++"] = standalone.standard_scores_dict(comparison)
return out
def test_on_random_sized():
data_obj = trivial_gen.balanced_random_size_gauss_blobs(
10.0, 2.0, 5, 100, 5, 0.5)
data_obj = partitioning.k_means(data_obj)
visu.plot_coords_label_color(data_obj)
visu.plot_centers_by_label_color(data_obj)
plt.show()
def croissants():
data = trivial_gen.croissants(100, 5.0, 0.5)
visu.plot_coords_label_color(data)
visu.plot_centers_by_label_color(data)
plt.show()
data = partitioning.k_means(data, initialisation="kmeans_++")
visu.plot_coords_label_color(data)
visu.plot_centers_by_label_color(data)
plt.show()
def test_rand_indexes_on_single_data(rand_test_data, initialisation, marker):
rand_indexes = []
for i in range(100):
data_obj = deepcopy(rand_test_data)
uncaught = True
while uncaught:
try:
data_obj = partitioning.k_means(data_obj,
initialisation=initialisation,
iterations=10)
rand_indexes.append(
comparative.scikit_rand_index(data_obj, rand_test_data))
uncaught = False
except ZeroDivisionError:
uncaught = True
rand_indexes.sort()
plt.plot(list(range(100)), rand_indexes, marker, label=initialisation)
def sample_process():
data_obj = trivial_gen.balanced_multiple_gauss_blobs(5.0, 2.0, 10, 2, 0.6)
_screenshot(data_obj)
partitioning.k_means(data_obj, iterations=3, on_step=_screenshot)
import py_osm_cluster.visualisation.visualisation as visu
import py_osm_cluster.visualisation.animation as anim
import matplotlib.pyplot as plt
parser = Parser("osm_data/krakow/krakow_center.osm")
all_buildings = parser.get_buildings_data_obj()
all_buildings.labels = [0 for i in all_buildings.coords]
churches = Coords()
for i in list(parser.ways.values()):
if "building" in i.tags and i.tags["building"] == "church":
churches.coords.append(list(i.geom.centroid.coords)[0])
churches.labels = [1 for i in churches.coords]
#print(churches)
#print(all_buildings)
all_buildings.c_number = len(churches.coords)
all_buildings.c_positions = churches.coords
all_buildings = partitioning.k_means(all_buildings,
iterations=1,
initialisation="default_pos")
print(all_buildings)
#all_buildings = partitioning.k_means(all_buildings,iterations=20,initialisation="kmeans_++")
#all_buildings = hierarchical.agglomerative(all_buildings,linkage="c-link")
visu.plot_coords_label_color(all_buildings)
visu.plot_coords_label_color(churches)
plt.show()
from py_osm_cluster.parser.parser import Parser
from py_osm_cluster.cluster import scikit
from py_osm_cluster.cluster import partitioning
from py_osm_cluster.eval import comparative as comparative
from py_osm_cluster.eval import standalone as standalone
from py_osm_cluster.cluster import hierarchical
import py_osm_cluster.visualisation.visualisation as visu
import py_osm_cluster.visualisation.animation as anim
import matplotlib.pyplot as plt
parser = Parser("osm_data/near_krakow/wegrzce_bibice.osm")
churches = Coords()
for i in list(parser.ways.values()):
if "building" in i.tags and i.tags["building"]=="church":
churches.coords.append(list(i.geom.centroid.coords)[0])
churches.labels = [1 for i in churches.coords]
data = parser.get_buildings_data_obj()
data.c_number = 10
data.labels =[0 for i in data.coords]
data = partitioning.k_means(data,initialisation="kmeans_++",iterations=25)
print(data)
visu.plot_coords_label_color(data)
visu.plot_coords_label_color(churches)
plt.show()
import py_osm_cluster.visualisation.visualisation as visu
import matplotlib.pyplot as plt
def pront(data_obj,name):
print("scikit_silhouette_score :"+str(standalone.scikit_silhouette_score(data_obj)))
print("scikit_calinski_harabaz_score :"+str(standalone.scikit_calinski_harabaz_score(data_obj)))
print("dunn_index :"+str(standalone.dunn_index(data_obj)))
print(result_to_string(standalone.general_evaluate_clustered_object(data_obj)))
visu.plot_coords_label_color(data_obj)
visu.plot_centers_by_label_color(data_obj)
plt.savefig(name+".png")
plt.close()
def result_to_string(dictionary):
out =""
for key in dictionary:
out = out + str(key) + "\n"
for key_2 in dictionary[key]:
out = out +" " +str(key_2) + ": " + str(dictionary[key][key_2]) +"\n"
return out
coords = trivial_gen.balanced_multiple_gauss_blobs(10.0,2.0,20,5,1.0)
pront(coords,"base")
print("!!!!!!!!!!!!!!!!!1")
newcoords = deepcopy(coords)
newcoords = partitioning.k_means(newcoords,initialisation="forgy")
pront(newcoords,"k_means")
from py_osm_cluster.util.coords import Coords as C
from py_osm_cluster.generator import trivial_gen as trivial_gen
from py_osm_cluster.eval import standalone as standalone
import py_osm_cluster.visualisation.visualisation as visu
import py_osm_cluster.visualisation.animation as anim
import matplotlib.pyplot as plt
from py_osm_cluster.cluster import partitioning
data_obj = C()
for i in range(3):
for j in range(3):
data_obj.coords.append((float(i), float(j)))
data_obj.labels.append(i)
data_obj.c_number = 3
data_2 = trivial_gen.multiple_gauss_blobs(10.0, 20, 5, 0.5)
print(standalone.dunn_index(data_2))
visu.plot_coords_label_color(data_2)
plt.show()
data_2 = partitioning.k_means(data_2)
print(standalone.dunn_index(data_2))
visu.plot_coords_label_color(data_2)
plt.show()