def pre_calendar(data_folder):
calendar = pd.read_csv(data_folder + '/calendar.csv')
#必要のない情報の削除
calendar.drop(['weekday', 'wday', 'month', 'year'], axis=1, inplace=True)
#カテゴリデータのの数値化
calendar = encoder_categorical(
calendar,
['event_name_1', 'event_type_1', 'event_name_2', 'event_type_2'])
return calendar
def pre_sales_train(data_folder):
sales_train = pd.read_csv(data_folder + '/sales_train_evaluation.csv')
sales_train = reduce_mem_usage(sales_train.iloc[:, :-28]) #容量減らす
sales_train = encoder_categorical(
sales_train, ['item_id', 'dept_id', 'cat_id', 'store_id', 'state_id'])
#製品情報"
product = sales_train.iloc[:, :6]
product['id'] = product['id'].str.replace('_evaluation', '_validation')
#売れ行き情報（目的変数）
demand = sales_train.iloc[:, 6:]
#売れ行き情報を減らす（直近２年分）
demand = demand.iloc[:, -730:]
return product, demand
def pre_calendar(data_folder):
calendar = pd.read_csv(data_folder + '/calendar.csv')
calendar.drop(['weekday', 'year'], axis=1, inplace=True)
calendar = encoder_categorical(
calendar,
['event_name_1', 'event_type_1', 'event_name_2', 'event_type_2'])
#dateの日付の部分を取得
d_day = pd.Series(calendar['date'].str[-2:], name='d_day', dtype='int')
calendar = pd.concat([calendar, d_day], axis=1)
#beggining of the year(boty)を作成
#month == 1かつd_dayがbotyリストの中にある場合、boty特徴量を1に
#beggining of the year
calendar['boty'] = 0
boty_list = [1, 2, 3, 4, 5, 6, 7]
for i in range(len(calendar)):
if (calendar.iat[i, 3] == 1) & (calendar.iat[i, 12] in boty_list):
calendar.iat[i, 13] = 1
#end of the year(eoty)
#month == 12かつd_dayがeotyリストにある数字と一緒だった場合eoty特徴量を1に
#end of the year
calendar['enty'] = 0
eoty_list = [25, 26, 27, 28, 29, 30, 31]
for i in range(len(calendar)):
if (calendar.iat[i, 3] == 12) & (calendar.iat[i, 12] in eoty_list):
calendar.iat[i, 13] = 1
#週末情報
#曜日の変数を担っているwdayが1か2だった場合、weekend変数を1に
#weekend
calendar['weekend'] = 0
for i in range(len(calendar)):
if (calendar.iat[i, 2] == 1) | (calendar.iat[i, 2] == 2):
calendar.iat[i, 14] = 1
return calendar
def pre_sell_prices(data_folder):
sell_prices = pd.read_csv(data_folder + '/sell_prices.csv')
sell_prices = encoder_categorical(sell_prices, ['item_id', 'store_id'])
return sell_prices