def label_transition_data(filename_to_label, **kwargs):
radius = kwargs['radius'] if 'radius' in kwargs else RADIUS
filename = kwargs['filename'] if 'filename' in kwargs else None
window = kwargs['window'] if 'window' in kwargs else WINDOW
n = kwargs['n'] if 'n' in kwargs else N
transition_probabilities = viterbi_training.run_viterbi(filename_to_label,filename=filename, radius=radius, window=window,n=n)
results = read_resulting_path('matched_files/Shopping2Rental_matched.csv')
if filename is not None:
with open(filename, 'w') as f:
f.write('Label, Distance Score, Backtrack Score \n')
for t, time in enumerate(transition_probabilities):
for seg1 in transition_probabilities[t]:
for seg2 in transition_probabilities[t][seg1]:
[way2_id,way2_index] = str(seg2).split(',')
start_node = get_node_id(int(way2_id), int(way2_index))
end_node = get_node_id(int(way2_id), int(way2_index)+1)
node_ids=(start_node[0],end_node[0])
node_ids2 = (end_node[0],start_node[0])
if ((node_ids == results[t]) or (node_ids2 == results[t])):
f.write('1, '+str(transition_probabilities[t][seg1][seg2][0]) + ', ' + str(transition_probabilities[t][seg1][seg2][1]) +'\n')
else:
f.write('-1, ' +str(transition_probabilities[t][seg1][seg2][0]) + ', ' + str(transition_probabilities[t][seg1][seg2][1]) +'\n')
return
return
def get_node_ids(matches):
node_ids = []
for i, match in enumerate(matches):
if match['way'] is None:
node_ids.append(None)
continue
# Don't query the same point twice
if i == 0 or match['way'] != matches[i-1]['way'] or match['index_of_segment'] != matches[i-1]['index_of_segment']:
start_node = get_node_id(match['way'], match['index_of_segment'])
start_node = re.findall(r'\d+', str(start_node))[0]
end_node = get_node_id(match['way'], match['index_of_segment'] + 1)
end_node = re.findall(r'\d+', str(end_node))[0]
node_ids.append((start_node, end_node))
return node_ids
def get_node_ids(matches):
node_ids = []
for i, match in enumerate(matches):
if match['way'] is None:
node_ids.append(None)
continue
# Don't query the same point twice
if i == 0 or match['way'] != matches[i - 1]['way'] or match[
'index_of_segment'] != matches[i - 1]['index_of_segment']:
start_node = get_node_id(match['way'], match['index_of_segment'])
start_node = re.findall(r'\d+', str(start_node))[0]
end_node = get_node_id(match['way'], match['index_of_segment'] + 1)
end_node = re.findall(r'\d+', str(end_node))[0]
node_ids.append((start_node, end_node))
return node_ids
def label_emission_data(file_to_label, **kwargs):
radius = kwargs['radius'] if 'radius' in kwargs else RADIUS
filename = kwargs['filename'] if 'filename' in kwargs else None
window = kwargs['window'] if 'window' in kwargs else WINDOW
n = kwargs['n'] if 'n' in kwargs else N
distance_only = kwargs[
'distance_only'] if 'distance_only' in kwargs else False
tmpfile = 'matched_files/Rental2Youssef_matched.csv'
#viterbi.run_viterbi(file_to_label,filename=tmpfile, radius=radius, window=window,n=n)
observations = read_observations(file_to_label)
results = read_resulting_path(tmpfile)
labeled = []
for t, obs in enumerate(observations):
possible_segments, prob, point = emission_probability.compute_emission_probabilities(
obs, radius, n)
for i, seg in enumerate(possible_segments):
start_node = get_node_id(seg['way_osm_id'], seg['index_in_way'])
end_node = get_node_id(seg['way_osm_id'], seg['index_in_way'] + 1)
node_ids = (start_node[0], end_node[0])
node_ids2 = (end_node[0], start_node[0])
if ((node_ids == results[t]) or (node_ids2 == results[t])):
if (distance_only):
labeled.append(seg['distance'])
else:
labeled.append((node_ids, seg['distance_score'],
seg['tangent_score'], 1))
else:
if (distance_only == False):
labeled.append((node_ids, seg['distance_score'],
seg['tangent_score'], -1))
if filename is not None:
with open(filename, 'w') as f:
f.write('Label, Distance Score, Tangent Score \n')
for l in labeled:
if (distance_only):
f.write(str(l) + '\n')
else:
f.write(
str(l[3]) + ', ' + str(l[1]) + ', ' + str(l[2]) + '\n')
return
return labeled
def label_transition_data(filename_to_label, **kwargs):
radius = kwargs['radius'] if 'radius' in kwargs else RADIUS
filename = kwargs['filename'] if 'filename' in kwargs else None
window = kwargs['window'] if 'window' in kwargs else WINDOW
n = kwargs['n'] if 'n' in kwargs else N
transition_probabilities = viterbi_training.run_viterbi(filename_to_label,
filename=filename,
radius=radius,
window=window,
n=n)
results = read_resulting_path('matched_files/Shopping2Rental_matched.csv')
if filename is not None:
with open(filename, 'w') as f:
f.write('Label, Distance Score, Backtrack Score \n')
for t, time in enumerate(transition_probabilities):
for seg1 in transition_probabilities[t]:
for seg2 in transition_probabilities[t][seg1]:
[way2_id, way2_index] = str(seg2).split(',')
start_node = get_node_id(int(way2_id), int(way2_index))
end_node = get_node_id(int(way2_id),
int(way2_index) + 1)
node_ids = (start_node[0], end_node[0])
node_ids2 = (end_node[0], start_node[0])
if ((node_ids == results[t])
or (node_ids2 == results[t])):
f.write('1, ' + str(transition_probabilities[t]
[seg1][seg2][0]) + ', ' +
str(transition_probabilities[t][seg1][seg2]
[1]) + '\n')
else:
f.write('-1, ' + str(transition_probabilities[t]
[seg1][seg2][0]) + ', ' +
str(transition_probabilities[t][seg1][seg2]
[1]) + '\n')
return
return
def label_emission_data(file_to_label, **kwargs):
radius = kwargs['radius'] if 'radius' in kwargs else RADIUS
filename = kwargs['filename'] if 'filename' in kwargs else None
window = kwargs['window'] if 'window' in kwargs else WINDOW
n = kwargs['n'] if 'n' in kwargs else N
distance_only = kwargs['distance_only'] if 'distance_only' in kwargs else False
tmpfile = 'matched_files/Rental2Youssef_matched.csv'
#viterbi.run_viterbi(file_to_label,filename=tmpfile, radius=radius, window=window,n=n)
observations = read_observations(file_to_label)
results = read_resulting_path(tmpfile)
labeled =[]
for t, obs in enumerate(observations):
possible_segments, prob, point = emission_probability.compute_emission_probabilities(obs,radius,n)
for i,seg in enumerate(possible_segments):
start_node = get_node_id(seg['way_osm_id'],seg['index_in_way'])
end_node = get_node_id(seg['way_osm_id'],seg['index_in_way']+1)
node_ids=(start_node[0],end_node[0])
node_ids2 = (end_node[0],start_node[0])
if ((node_ids == results[t]) or (node_ids2 == results[t])):
if (distance_only):
labeled.append(seg['distance'])
else:
labeled.append((node_ids,seg['distance_score'],seg['tangent_score'], 1))
else:
if (distance_only == False):
labeled.append((node_ids,seg['distance_score'],seg['tangent_score'], -1))
if filename is not None:
with open(filename, 'w') as f:
f.write('Label, Distance Score, Tangent Score \n')
for l in labeled:
if (distance_only):
f.write(str(l) + '\n')
else:
f.write(str(l[3]) + ', ' + str(l[1]) +', '+str(l[2]) + '\n')
return
return labeled