def _preprocess_order_data(self, order, need_unitize=True):
order = order.loc[order.first_cate_code.isin(self._all_cate_codes)]
order = order.loc[order.order_date >= '2015-09-01']
order = order.loc[~order.customer_code.str.contains('\\N', regex=False
)]
removed_column_names = [
'org_code', 'region_code', 'region_name', 'road',
'fin_segment1_code', 'fin_segment1_name', 'fin_segment2_code',
'fin_segment2_name', 'received_qty', 'return_qty'
]
order.drop(columns=removed_column_names, inplace=True)
order = order.sort_values(by='order_date').reset_index(drop=True)
str_column_names = ['sales_cen_code', 'customer_code', 'item_code']
remove_whitespace(order, str_column_names)
order['district'] = order.district.str.replace(r'\\N', '未知')
order['channel_name'] = order.channel_name.apply(
lambda x: transform_channel(x))
order['channel_name'] = order.channel_name.str.replace(r'\\N', '未知')
order['sales_chan_name'] = order.sales_chan_name.str.replace(
r'\\N', '未知')
order.project_flag.fillna('未知', inplace=True)
order['project_flag'] = order.project_flag.apply(
lambda x: transform_project_flag(x))
if need_unitize: # 是否需要单位化，即以万作为单位
order['ord_qty'] = order.ord_qty / 10000
order['ord_amount'] = order.ord_amount / 10000
return order
def _preprocess_dis_data(self, dis, need_unitize=True):
dis.drop(columns=[
'bu_code', 'bu_name', 'region_code', 'region_name', 'road'
],
inplace=True)
dis = dis.loc[dis.first_cate_code.isin(self._all_cate_codes)]
dis = dis.loc[dis.customer_code.str.startswith('C')]
dis = dis.sort_values(by='order_date').reset_index(drop=True)
dis['sales_cen_code'] = dis.sales_cen_code.astype(str).apply(
lambda x: x.strip())
dis['customer_code'] = dis.customer_code.astype(str).apply(
lambda x: x.strip())
dis['item_code'] = dis.item_code.astype(str).apply(lambda x: x.strip())
dis['district'] = dis.district.str.replace(r'\\N', '未知')
dis['customer_type'] = dis.customer_type.str.replace(r'\\N', '未知')
dis['is_usable'] = dis.is_usable.astype(str)
dis['is_usable'] = dis.is_usable.str.replace(r'\\N', '未知')
dis['channel_name'] = dis.channel_name.apply(
lambda x: transform_channel(x))
dis['dis_qty'] = np.round(dis.dis_qty).astype(int)
dis['item_price'] = dis.item_price.astype(str).str.replace(
r'\\N', '-1.0').astype(float)
dis['dis_amount'] = dis.dis_qty * dis.item_price
if need_unitize:
dis['dis_qty'] = dis.dis_qty / 10000
dis['dis_amount'] = dis.dis_amount / 10000
return dis
def _preprocess_inv_data(self, inv, need_unitize=True):
inv = inv.loc[inv.first_cate_code.isin(self._all_cate_codes)]
inv = inv.sort_values(by='order_date').reset_index(drop=True)
if need_unitize: # 是否需要单位化，即以万作为单位
inv['inv_qty'] = inv.inv_qty / 10000
inv['inv_amount'] = inv.inv_amount / 10000
inv['district'] = inv.district.str.replace(r'\\N', '未知')
inv['channel_name'] = inv.channel_name.apply(
lambda x: transform_channel(x))
inv['sales_chan_name'] = inv.sales_chan_name.str.replace(r'\\N', '未知')
return inv
def _preprocess_inv_data(self, inv, need_unitize=True):
inv.drop(columns=['bu_code',
'bu_name',
'region_code',
'region_name'], inplace=True)
inv = inv.loc[inv.first_cate_code.isin(self._all_cate_codes)]
inv = inv.sort_values(by='order_date').reset_index(drop=True)
inv['sales_cen_code'] = inv.sales_cen_code.astype(str).apply(lambda x: x.strip())
inv['customer_code'] = inv.customer_code.astype(str).apply(lambda x: x.strip())
inv['item_code'] = inv.item_code.astype(str).apply(lambda x: x.strip())
inv['district'] = inv.district.str.replace(r'\\N', '未知')
inv['channel_name'] = inv.channel_name.apply(lambda x: transform_channel(x))
inv['sales_chan_name'] = inv.sales_chan_name.str.replace(r'\\N', '未知')
if need_unitize:
inv['inv_qty'] = inv.inv_qty / 10000
inv['inv_amount'] = inv.inv_amount / 10000
return inv