def get_testset_item():
raw_conf = get_raw_conf()
db_conf = get_db_conf()
print "start to connect the db.."
try:
conn = mysql.connector.connect(host=db_conf['host'],
user=db_conf['user'],
passwd=db_conf['passwd'])
except Exception as err:
logging.error(err)
cur = conn.cursor()
cur.execute('use %s' % (db_conf['db_name']))
print 'start to search the users..'
sql_str = 'select distinct user from user_train'
cur.execute(sql_str)
user_list = cur.fetchall()
dates = datetime.date(2014, 12, 19)
infile = open('../data/item_train.csv', 'rb')
item_dict = {}
print 'start to get the items.'
for idx, row in enumerate(DictReader(infile)):
if row['item_id'] not in item_dict:
item_dict[row['item_id']] = row['item_category']
else:
print row['item_id']
# TODO user_list 和 item_dict 可以存起来的
print 'got the users items'
for user in user_list:
for item in item_dict:
yield [user, item, item_dict[item]]
feature抽取
'''
import mysql.connector
import sys
sys.path.append('..')
import logging, time
from util.db import mydb
from util.item import item
from util.get_item import get_item, get_raw_conf, get_db_conf
import datetime
import csv
import itertools
cf = get_db_conf()
raw_data = get_raw_conf()
def get_mydb():
host = cf["host"]
user = cf["user"]
passwd = cf["passwd"]
if passwd == "null":
passwd = ""
mdb = mydb(host=host, user=user, passwd=passwd, port=3306)
return mdb
# 将所有输出数据或者配置放到一个文件里还是一个不错的选择的
# 这样就不用在各个文件中添加read_conf了
import sys
sys.path.append('..')
import mysql.connector
import logging
from util.item import item
from util.get_item import get_db_conf, get_one_item, get_raw_conf
raw_conf = get_raw_conf()
data_path = raw_conf['u_tr_time']
dbconf = get_db_conf()
host = dbconf['host']
user = dbconf['user']
passwd = dbconf['passwd']
if passwd == 'null':
passwd = ""
conn = mysql.connector.connect(host=host, user=user, passwd=passwd)
def create():
cur = conn.cursor()