def get_seqs_by_weekdays(fp):
result = []
dt_locs = get_wifi_seqs(fp, 60*10, 20)
days = ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday']
for day in days:
#print day
#pprint(locs)
seqs = []
seq_freq = {}
avg_seq_len = 0
for dt, entries in dt_locs:
dt_obj = datetime.strptime(dt, "%d%b%Y")
weekday = dt_obj.strftime("%A")
#if weekday != 'Sunday' and weekday != 'Saturday':
if weekday == day:
seqs.append(entries)
#pprint(seqs)
result.append((day, seqs))
for pair in result:
print pair[0]
for entry in pair[1]:
print entry
return result
def get_feature():
id_feature = {}
addr_dir = r'C:\Users\Sophie\workspace\Personality\old_code\data\by_subjects'
for file in os.listdir(addr_dir):
if not file.endswith('.csv'):
continue
id = file.split('.')[0][-2:]
if int(id) >= 45:
continue
# if id in WIFI_OFF_CAMPUS:
# continue
fp = os.path.join(addr_dir, file)
print '----------'
print 'id: ' + id
#seqs = get_wifi_seqs(fp, 60*10, 20)
seqs = get_wifi_seqs(fp, 60*5, 30)
#print seqs
#print len(seqs)
result = get_avg_edit_dist(seqs)
print result
id_feature[id] = result
return id_feature
def get_feature():
id_feature = {}
addr_dir = os.path.join(CUR_DIR, 'data', 'by_subjects')
for file in os.listdir(addr_dir):
if not file.endswith('.csv'):
continue
id = file.split('.')[0][-2:]
if int(id) >= 45:
continue
if id in WIFI_OFF_CAMPUS:
continue
fp = os.path.join(addr_dir, file)
print '----------'
print 'id: ' + id
seqs = get_wifi_seqs(fp, 60*10, 30)
#result = avg_seq_len(seqs)
result = num_pattern(seqs)
print result
id_feature[id] = result
return id_feature
ids = []
input_dir = os.path.join(CUR_DIR, 'data', 'by_subjects')
for file in os.listdir(input_dir):
if not file.endswith('.csv'):
continue
id = file.split('.')[0][-2:]
if int(id) >= 45:
continue
if id in WIFI_OFF_CAMPUS:
continue
fp = os.path.join(input_dir, file)
ids.append(id)
print id
seqs = get_wifi_seqs(fp, DURATION_CUT, NUM_DAYS)
# for seq in seqs:
# print seq
replace_home(seqs, id)
seqs_by_subject.append(seqs)
all_seqs.extend(seqs)
freq_pat = []
gsp(all_seqs, 1, [], freq_pat)
#pprint(freq_pat)
print len(freq_pat)
for i, p in enumerate(freq_pat):
print i, p