traj_list = io.load("1_traj_seg.dt")
traj_list = traj_list[:1000]
X = np.fromfile("gaussian_representation.dat", dtype=float)
D = io.load_distance_matrix("distance1.npz")
no_of_cluster = 12
gmm = GMM(n_components=no_of_cluster, n_iter=1000)
labels = gmm.fit_predict(X)
# Postprocessing
clusters = [[] for i in range(no_of_cluster)]
no = len(traj_list)
for i in range(no):
label = int(labels[i])
clusters[label].append(traj_list[i])
silhoutte_score = metrics.silhouette_score(D, labels, sample_size=1000)
print("Silhoutte Coefficient : %.3f" % silhoutte_score)
# Plotting Clustered Trajectories
color_list = plt.rcParams['axes.prop_cycle'].by_key()['color']
for i in range(no_of_cluster):
for traj in clusters[i]:
next_color = color_list[i % len(color_list)]
tplot.plot_traj(traj, next_color, alpha=1)
tplot.plot_map()
if not flied:
suspi = path
traj_list.append(path)
print(countFlied)
print(countFlied2)
'''
for traj in traj_list:
tplot.plot_traj(traj)
'''
print(suspi)
tplot.plot_traj(suspi)
tplot.plot_map()
exit(0)
filename1 = "3.traj"
filename2 = "6.traj"
d1 = io.load(filename1)
d2 = io.load(filename2)
key1 = set(d1.keys())
key2 = set(d2.keys())
common = key1.intersection(key2)
for i in range(no_of_cluster):
traj_list = generate_cluster(no_of_traj_each_cluster, traj_length,
noise)
normal_traj_list += traj_list
outlier_traj_list = []
for i in range(no_of_outlier):
outlier_traj_list += generate_cluster(1, traj_length, noise)
# Plotting Generated Trajectories
plt.figure(1)
plt.subplot(121)
for traj in normal_traj_list:
aplt.plot_traj(traj)
for traj in outlier_traj_list:
aplt.plot_traj(traj, "r")
# Precomputation
traj_list = normal_traj_list + outlier_traj_list
random.shuffle(traj_list)
# Calculating Distance Matrix
s.start()
D = calculate_distance_matrix(traj_list, eps)
s.stop("Distance matrix calculated")
# Clustering