info = class_info[item]
loc = info['location']
periods = info['periods']
# organize by weekday
for period in periods:
wkd = to_weekday[period['day']]
if not wkd in schedule:
schedule[wkd] = []
start = period['start']
end = period['end']
schedule[wkd].append((loc, start, end))
# sort by start time
for wkd in schedule:
schedule[wkd].sort(key=lambda tup:tup[1])
#pp.pprint(schedule)
result = get_late_time(id, schedule)
feature = get_ratio(result)
#feature = get_late_var(result)
id_feature[id] = feature
return id_feature
if __name__ == '__main__':
id_features = get_feature()
fill_miss_values(id_features, 3, ['00', '36', '39', '56'])
write_multi_features_to_csv(id_features, ['early', 'late', 'absent'])
#fill_miss_values(id_features, 1, ['00', '36', '39', '56'])
#write_feature_to_csv(id_features, 'late_time_var')
def get_feature(func):
id_feature = {}
for file in os.listdir(dir):
if not file.endswith('.txt') :
continue
id = file.split('.')[0][-2:]
if id in ['20', '36', '42', '47', '43']:
continue
print '--------'
print 'id: ' + id
result = func(id)
id_feature[id] = result
return id_feature
if __name__ == '__main__':
#id_feature = get_feature(loc_entropy)
#write_feature_to_csv(id_feature, 'dining_loc_entropy', False)
# id_features = get_feature(meal_rate)
# write_multi_features_to_csv(id_features, ['breakfast','lunch','supper','snack'], False)
id_features = get_feature(time_var)
write_multi_features_to_csv(id_features, ['breakfast_var','lunch_var','supper_var','snack_var'])
