def detect_daily_work_office(user_id,day):
# say should be from 1 to 5
Sections=get_section_db()
office_candidate=[]
home=detect_home(user_id)
if home == 'N/A':
return 'N/A'
# print(list_first_pnt)
for section in Sections.find({"$and":[{"user_id": user_id},{ "section_start_point": { "$ne": None }}]}):
section_start_pnt=section['section_start_point']
section_end_pnt=section['section_end_point']
if Is_date(section['section_start_time'],day)==True:
# parameter that the distance away from home:
away_home=400
if not Is_place(section_start_pnt,home,away_home) and not Is_place(section_end_pnt,home,away_home):
office_candidate.append(section['track_points'][0])
office_candidate.append(section['track_points'][-1])
elif Is_place(section_start_pnt,home,away_home) and not Is_place(section_end_pnt,home,away_home):
office_candidate.append(section['track_points'][-1])
elif not Is_place(section_start_pnt,home,away_home) and Is_place(section_end_pnt,home,away_home):
office_candidate.append(section['track_points'][0])
if len(office_candidate)>0:
office_candidate = sorted(office_candidate, key=lambda k: parser.parse(k['time']))
# print(office_candidate)
weighted_office_candidate=get_static_pnts(office_candidate)
office_location=most_common(weighted_office_candidate,200)
# print(len(office_candidate))
# print(len(weighted_office_candidate))
# print(calculate_appearance_rate(office_candidate,office_location))
# print(calculate_appearance_rate(weighted_office_candidate,office_location))
return office_location
else:
return 'N/A'
def detect_home_2(user_id):
Sections = get_section_db()
list_first_pnt = []
list_home_candidate = []
for section in Sections.find({
'$and': [{
'user_id': user_id
}, {
'section_start_point': {
'$ne': None
}
}]
}):
list_first_pnt.append(section['track_points'][0])
list_home_candidate = get_first_daily_point(list_first_pnt)
if len(list_home_candidate) == 0:
return 'N/A'
list_home_candidate_cood = []
for pnt in list_home_candidate:
list_home_candidate_cood.append(pnt['track_location']['coordinates'])
list_home_candidate_np = np.asarray(list_home_candidate_cood)
minlat = np.min(list_home_candidate_np[:, 0])
minlng = np.min(list_home_candidate_np[:, 1])
list_home_candidate_np_2 = np.zeros((len(list_home_candidate_np), 2))
list_home_candidate_np_2[:, 0] = (list_home_candidate_np[:, 0] -
minlat) * 89.7
list_home_candidate_np_2[:, 1] = (list_home_candidate_np[:, 1] -
minlng) * 112.7
db = DBSCAN(eps=0.2, min_samples=3)
db_fit = db.fit(list_home_candidate_np_2)
db_labels = db_fit.labels_
new_db_labels = db_labels[db_labels != -1]
if new_db_labels != []:
frequency = Counter(new_db_labels)
max_fre = frequency.most_common()[0][1]
new_location = list_home_candidate_np[db_labels != -1]
label_unique = np.unique(new_db_labels)
cluster_center = np.zeros((len(label_unique), 3))
home_list = []
for label in label_unique:
sub_location = new_location[new_db_labels == label]
if len(sub_location) / max_fre >= 0.3:
temp_center = np.mean(sub_location, axis=0)
home_list.append(list(temp_center))
return home_list
else:
return [most_common(list_home_candidate, 200)]
def detect_home(user_id):
Sections=get_section_db()
list_first_pnt=[]
list_home_candidate=[]
for section in Sections.find({"$and":[{"user_id": user_id},{ "section_start_point": { "$ne": None }}]}):
list_first_pnt.append(section['track_points'][0])
# print(list_first_pnt)
list_home_candidate=get_first_daily_point(list_first_pnt)
# print(type(list_home_candidate))
# print(type(list_home_candidate[0]))
if len(list_home_candidate)==0:
return 'N/A'
else:
home_location=most_common(list_home_candidate,500)
return home_location
def json_to_records(js):
"""Extract records from this json object by finding the leaf nodes and taking the nodes with most common counts as the columns
"""
records = []
counter = json_counter(js)
if counter:
num_entries = common.most_common(
[len(values) for values in counter.values()])
# get the fields with the correct number of entries
fields = [
key for key in counter.keys() if len(counter[key]) == num_entries
]
for i in range(num_entries):
result = dict([(field, counter[field][i]) for field in fields])
records.append(result)
return records
def detect_home_2(user_id):
Sections = get_section_db()
list_first_pnt = []
list_home_candidate = []
for section in Sections.find({'$and': [{'user_id': user_id}, {'section_start_point': {'$ne': None}}]}):
list_first_pnt.append(section['track_points'][0])
list_home_candidate = get_first_daily_point(list_first_pnt)
if len(list_home_candidate) == 0:
return 'N/A'
list_home_candidate_cood = []
for pnt in list_home_candidate:
list_home_candidate_cood.append(pnt['track_location']['coordinates'])
list_home_candidate_np = np.asarray(list_home_candidate_cood)
minlat = np.min(list_home_candidate_np[:, 1])
minlng = np.min(list_home_candidate_np[:, 0])
list_home_candidate_np_2 = np.zeros((len(list_home_candidate_np), 2))
list_home_candidate_np_2[:, 1] = (list_home_candidate_np[:, 0] - minlat) * 89.7
list_home_candidate_np_2[:, 0] = (list_home_candidate_np[:, 1] - minlng) * 112.7
db = DBSCAN(eps=0.2, min_samples=3)
db_fit = db.fit(list_home_candidate_np_2)
db_labels = db_fit.labels_
new_db_labels = db_labels[db_labels != -1]
if new_db_labels != []:
frequency = Counter(new_db_labels)
max_fre = frequency.most_common()[0][1]
new_location = list_home_candidate_np[db_labels != -1]
label_unique = np.unique(new_db_labels)
cluster_center = np.zeros((len(label_unique), 3))
home_list = []
for label in label_unique:
sub_location = new_location[new_db_labels == label]
if len(sub_location) / max_fre >= 0.3:
temp_center = np.mean(sub_location, axis=0)
home_list.append(list(temp_center))
return home_list
else:
return [most_common(list_home_candidate, 200)]