/
order.py
181 lines (161 loc) · 8.07 KB
/
order.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
import os.path
import numpy as np
from itertools import permutations
from sklearn.metrics import pairwise_distances, jaccard_score
from pyjarowinkler import distance
from util import *
from k_opt_tsp import tsp_2_opt
def get_floor_order_str(floors):
id_list = list(set(range(len(floors))))
first_id = np.min(id_list)
id_str_list = [str(i-first_id) for i in id_list]
return "".join(id_str_list)
def id_str_order(floors):
id_list = list(set(range(len(floors))))
first_id = np.min(id_list)
id_str_list = [str(i-first_id) for i in id_list]
return "".join(id_str_list), "".join(id_str_list[::-1])
def floor_str_order(floors):
floor_id = []
for floor in floors:
floor_id.append(floor_map[floor])
first_id = np.min(floor_id)
sorted_ids = [i-first_id for i in floor_id]
sorted_ids.sort()
floor_ids = [str(sorted_ids.index(i-first_id)) for i in floor_id]
return "".join(floor_ids)
def hit(t_str, f_str):
return t_str==f_str
def jaccard_distance(cluster_i, cluster_j):
return jaccard_score(np.sum(cluster_i, axis = 0)!=0, np.sum(cluster_j, axis = 0)!=0)
def adapted_jaccard_distance(cluster_i, cluster_j):
cluster_i_e = []
cluster_j_e = []
for cluster in np.sum(cluster_i, axis = 0):
if cluster!=0:
cluster_i_e.append(cluster)
for cluster in np.sum(cluster_j, axis = 0):
if cluster!=0:
cluster_j_e.append(cluster)
f_share = np.dot(np.sum(cluster_i, axis = 0), np.sum(cluster_j, axis = 0))
f_diff_i = np.dot(np.sum(cluster_i, axis = 0)==0, np.sum(cluster_j, axis = 0))*np.average(cluster_i_e)
f_diff_j = np.dot(np.sum(cluster_j, axis = 0)==0, np.sum(cluster_i, axis = 0))*np.average(cluster_j_e)
f_diff = f_diff_i+f_diff_j
return f_share/(f_share+f_diff)
def calculate_dist_statistics(dists):
return {
# "min": np.min(dists[dists != 0]),
"min": np.min(np.min(dists),0),
"max": np.max(dists),
"median": np.median(dists),
"mean": np.mean(dists)
}
def show_cluster_pair_statistics(cluster_i, cluster_j, args):
if args.order_id==0: # pairwise
inter_dists = pairwise_distances(cluster_i, cluster_j)
elif args.order_id==1: # AP, jaccard
inter_dists = jaccard_distance(cluster_i, cluster_j)
else: # AP, adapted jaccard
inter_dists = adapted_jaccard_distance(cluster_i, cluster_j)
inter_stats = calculate_dist_statistics(inter_dists)
return inter_stats
def show_cluster_statistics(cluster, args):
return show_cluster_pair_statistics(cluster, cluster, args)
def show_result(dataset, floors, floor_labels, args, building):
# intra-cluster
# for tmp_floor in floors:
# tmp_floor_dataset = [d for d, l in zip(dataset, floor_labels) if l == tmp_floor]
# print(show_cluster_statistics(tmp_floor_dataset, args))
# inter-cluster
selected_dist_matrix = np.zeros((len(floors), len(floors)))
if args.order_id==0: # pairwise
for i, floor_i in enumerate(floors):
floor_i_dataset = [d for d, l in zip(dataset, floor_labels) if l == floor_i]
for j in range(i+1, len(floors)):
floor_j = floors[j]
# print(f"Floor {floor_i} and Floor {floor_j}:")
floor_j_dataset = [d for d, l in zip(dataset, floor_labels) if l == floor_j]
inter_stats = show_cluster_pair_statistics(floor_i_dataset, floor_j_dataset, args)
selected_dist_matrix[i,j] = inter_stats["median"]
selected_dist_matrix[j,i] = inter_stats["median"]
# print(inter_stats)
else: # jaccard
for i, floor_i in enumerate(floors):
floor_i_dataset = [d for d, l in zip(get_ap_obs(args), floor_labels) if l == floor_i]
for j in range(i+1, len(floors)):
floor_j = floors[j]
# print(f"Floor {floor_i} and Floor {floor_j}:")
floor_j_dataset = [d for d, l in zip(get_ap_obs(args), floor_labels) if l == floor_j]
inter_stats = show_cluster_pair_statistics(floor_i_dataset, floor_j_dataset, args)
selected_dist_matrix[i,j] = inter_stats["median"]
selected_dist_matrix[j,i] = inter_stats["median"]
# print(selected_dist_matrix)
dists = []
perms = []
keys = set()
topk = args.k
for perm in permutations(range(len(floors))):
perm_key = "".join([str(item) for item in perm])
reverse_key = "".join([str(item) for item in perm[::-1]])
if perm_key in keys or reverse_key in keys:
continue
sum_dist = 0
for (i,j) in zip(perm[:-1], perm[1:]):
sum_dist += selected_dist_matrix[i,j]
dists.append(sum_dist)
perms.append(perm)
keys.add(perm_key)
o1, o2 = id_str_order(floors)
if args.order_id==0: # pairwise, want minimum
sort_indicies = np.argsort(dists)
for i in range(len(sort_indicies)):
floor_order = [floors[k] for k in perms[sort_indicies[i]]]
order_dist = dists[sort_indicies[i]]
d1 = distance.get_jaro_distance(o1, floor_str_order(floor_order), winkler=True, scaling=0.1)
d2 = distance.get_jaro_distance(o2, floor_str_order(floor_order), winkler=True, scaling=0.1)
# print(f"perm detected: {floor_order}, with min_dist: {order_dist}")
if i<topk:
print(f"perm detected: {floor_order}, with edit distance from true order: {max(d1, d2)}")
if hit(o1, floor_str_order(floor_order)) or hit(o2, floor_str_order(floor_order)):
print("correct sequence detected at {}".format(i+1))
break
else: # jaccard, want maximum
sort_indicies = np.flipud(np.argsort(dists))
for i in range(len(sort_indicies)):
floor_order = [floors[k] for k in perms[sort_indicies[i]]]
order_dist = dists[sort_indicies[i]]
d1 = distance.get_jaro_distance(o1, floor_str_order(floor_order), winkler=True, scaling=0.1)
d2 = distance.get_jaro_distance(o2, floor_str_order(floor_order), winkler=True, scaling=0.1)
if i<topk:
print(f"perm detected: {floor_order}, with edit distance from true order: {max(d1, d2)}")
if hit(o1, floor_str_order(floor_order)) or hit(o2, floor_str_order(floor_order)):
print("correct sequence detected at {}".format(i+1))
break
def show_result_2_opt(dataset, floors, floor_labels, args, building):
selected_dist_matrix = np.zeros((len(floors), len(floors)))
if args.order_id == 0: # pairwise
for i, floor_i in enumerate(floors):
floor_i_dataset = [d for d, l in zip(dataset, floor_labels) if l == floor_i]
for j in range(i + 1, len(floors)):
floor_j = floors[j]
# print(f"Floor {floor_i} and Floor {floor_j}:")
floor_j_dataset = [d for d, l in zip(dataset, floor_labels) if l == floor_j]
inter_stats = show_cluster_pair_statistics(floor_i_dataset, floor_j_dataset, args)
selected_dist_matrix[i, j] = inter_stats["median"]
selected_dist_matrix[j, i] = inter_stats["median"]
# print(inter_stats)
else: # jaccard
for i, floor_i in enumerate(floors):
floor_i_dataset = [d for d, l in zip(get_ap_obs(args), floor_labels) if l == floor_i]
for j in range(i + 1, len(floors)):
floor_j = floors[j]
floor_j_dataset = [d for d, l in zip(get_ap_obs(args), floor_labels) if l == floor_j]
inter_stats = show_cluster_pair_statistics(floor_i_dataset, floor_j_dataset, args)
selected_dist_matrix[i, j] = 1 / (inter_stats["median"])
selected_dist_matrix[j, i] = 1 / (inter_stats["median"])
order_2_opt = tsp_2_opt(selected_dist_matrix, list(range(len(floors))))
floor_pred = "".join([str(item) for item in order_2_opt])
o1 = get_floor_order_str(floors)
dist = distance.get_jaro_distance(o1, floor_pred, winkler=True, scaling=0.1)
print (f"edit distance: {dist}")
return order_2_opt