def data_split(self):
"""
训练集数据切分为训练集及验证集。
validationPids：需要被选做为验证集的司机id
"""
self.drivers_pd = pd.read_csv(self.root_path +
"drivers_img_nop081_list.csv")
self.imgs_pd = self.drivers_pd["img"]
self.class_pd = self.drivers_pd["classname"]
self.subject_pd = self.drivers_pd["subject"]
self.choices = ["p035", "p047"]
DataSplit.split(
choice_ids=self.choices,
train_pd_path=self.root_path + "drivers_img_nop081_list.csv",
train_aug_pd_path=self.root_path + "drivers_img_aug_list.csv",
train_dir=self.train_dir,
val_dir=self.val_dir,
test_dir=self.test_dir,
origin_test_dir=self.root_path + "imgs/test",
saved_weights_dir=self.saved_weights)
return self
user_item_map[words[0]][words[1]] += 1
count_map = {}
count_user_item_map = {}
for user in user_item_map:
for item, cnt in user_item_map[user].items():
count_map.setdefault(cnt, 0)
count_map[cnt] += 1
count_user_item_map.setdefault(cnt, {})
count_user_item_map[cnt].setdefault(user, set())
count_user_item_map[cnt][user].add(item)
# for cnt in count_map:
# print 'the number of behavior which occurs %d is %d' %(cnt, count_map[cnt])
return count_map, count_user_item_map
if __name__ == '__main__':
id_position_map = DataSplit.itemIDMap('../File/tianchi_mobile_recommend_train_item.csv')
test_user_item_map, test_user_item_map_size = DataSplit.userOperate('../File/user_test_data.csv', 4, id_position_map)
fp_behavior_occur_count = open('../File/AnalysisResult/behavior_occur_count.csv', 'w')
total1 = 0
for i in xrange(1, 5):
fp_behavior_occur_count.write('behavior type is %d: \n' %i)
count_map, count_user_item_map = userBuyAppetite('../File/user_train_data.csv', i, id_position_map)
total = 0
for cnt in count_map:
# print count_map[cnt], len(count_user_item_map[cnt])
bingos, test_set_num = DataSplit.computeRatio1(count_user_item_map[cnt], test_user_item_map)
fp_behavior_occur_count.write('behavior occurs = %d, in train data(%d,%d,%.2lf), in test data(%d,%d,%.2lf)\n' \
%(cnt, bingos, count_map[cnt], 100.0 * bingos / count_map[cnt], bingos, test_set_num, 100.0 * bingos / test_set_num))
total += bingos
# total1 += len(count_user_item_map[cnt])
fp_behavior_occur_count.write('%d\n' %total)
count_map = {}
count_user_item_map = {}
for user in user_item_map:
for item, cnt in user_item_map[user].items():
count_map.setdefault(cnt, 0)
count_map[cnt] += 1
count_user_item_map.setdefault(cnt, {})
count_user_item_map[cnt].setdefault(user, set())
count_user_item_map[cnt][user].add(item)
# for cnt in count_map:
# print 'the number of behavior which occurs %d is %d' %(cnt, count_map[cnt])
return count_map, count_user_item_map
if __name__ == '__main__':
id_position_map = DataSplit.itemIDMap(
'../File/tianchi_mobile_recommend_train_item.csv')
test_user_item_map, test_user_item_map_size = DataSplit.userOperate(
'../File/user_test_data.csv', 4, id_position_map)
fp_behavior_occur_count = open(
'../File/AnalysisResult/behavior_occur_count.csv', 'w')
total1 = 0
for i in xrange(1, 5):
fp_behavior_occur_count.write('behavior type is %d: \n' % i)
count_map, count_user_item_map = userBuyAppetite(
'../File/user_train_data.csv', i, id_position_map)
total = 0
for cnt in count_map:
# print count_map[cnt], len(count_user_item_map[cnt])
bingos, test_set_num = DataSplit.computeRatio1(
count_user_item_map[cnt], test_user_item_map)
fp_behavior_occur_count.write('behavior occurs = %d, in train data(%d,%d,%.2lf), in test data(%d,%d,%.2lf)\n' \
time = words[5]
year1, month1, day1, hour1 = timeSparse(time)
date1 = datetime.datetime(year1, month1, day1)
interval = int((date - date1).days)
# if file_point_map.has_key(interval) == False:
# output_file = '%02d.csv' %interval
# fp_output = open(output_file, 'w')
# file_point_map[interval] = fp_output
file_point_map[interval].write('%s' %line)
if __name__ == '__main__':
#分割文件为31个文件，每天一个文件
<<<<<<< HEAD
# dataSplitByDay('../File/tianchi_mobile_recommend_train_user.csv')
# dataSplitByDay('../File/complete_addDistance_train_user.csv')
id_position_map = DataSplit.itemIDMap('../File/tianchi_mobile_recommend_train_item.csv')
=======
dataSplitByDay('../File/tianchi_mobile_recommend_train_user.csv')
# dataSplitByDay('../File/complete_addDistance_train_user.csv')
# id_position_map = DataSplit.itemIDMap('../File/tianchi_mobile_recommend_train_item.csv')
>>>>>>> 972f41ad2af1fc238e65d04b58a152301666f6bf
# id_position_map = DataSplit.itemIDMap('../File/train_user_complete_by_userInfo.csv')
"""
""分析每天的user-item行为
"""
import os
total_user_item_behavior_list = [[0 for i in xrange(0, 8)] for i in xrange(0, 5)]
rootDir = '../File/EverydayData/'
for cnt in xrange(0, 1):
print 'process %02d' %cnt
tag = 0