def func2():
data = pd.read_csv('./data/train.csv')
data = data[['user_id','brand_id','type','month','day']]
data = np.array(data)
user_id = func.read('user_id')
item_id = func.read('item_id')
user_item_time = func.build_user_item_time(data,user_id)
item_time = itemmodel.build_item_time(data,item_id)
item_factor = itemmodel.cal_item_time_factor(item_time)
func.write(user_item_time,'user_item_time')
func.write(item_time,'item_time')
return (user_item_time,item_time,item_factor)
def func3():
train,test = func.divide_data()
user_id = func.read('user_id')
item_id = func.read('item_id')
train_u_i_t = func.build_user_item_time(train,user_id)
#建立测试集的用户-物品词典，这里不需要时间序列
test_u_i = func.build_user_item(test,user_id)
#建立训练集的物品-时间词典
train_item_time = itemmodel.build_item_time(train,item_id)
#建立物品的时间衰减因子
item_factor = itemmodel.cal_item_time_factor(train_item_time)
func.write(train_u_i_t,'train_user_item_time')
func.write(test_u_i,'test_user_item')
func.write(train_item_time,'train_item_time')
return (train_u_i_t,test_u_i,train_item_time,item_factor)