logistics/产品上限优化.ipynb

In [ ]:

import pandas as pd
from utils.gtools import MySQLconnect
categories = '94 - Office Desks'
with MySQLconnect('ods') as db:
    engine = db.engine()
    sql = f"""
WITH a AS (
	SELECT
		t1.SPU,
		t2.SKU,
		t1.产品分类,
		t1.添加时间,
		order_date,
		t2.成本价,
		opl.product_price_dollar,
		opl.product_num,
		TIMESTAMPDIFF( MONTH, t1.添加时间, order_date ) AS month_diff 
	FROM
		ods.stg_bayshop_litfad_spu t1
		LEFT JOIN ods.stg_bayshop_litfad_sku t2 ON t2.产品PID = t1.产品PID
		LEFT JOIN dws.order_product_list opl ON t2.SKU = opl.SKU 
	WHERE
		t1.添加时间 BETWEEN '2023-01-01' 
		AND '2024-12-31 23:59:59' 
		AND 产品分类 = '{categories}'
		AND t2.SKU IS NOT NULL 
	),
	b AS (
	SELECT
		SPU,
		SKU,添加时间,产品分类,成本价,
		b.erp_package_vol,
		order_date,
	IF
		( month_diff >= 6, NULL, month_diff ) AS month_diff,
		ROW_NUMBER() over ( PARTITION BY SKU ORDER BY order_date DESC ) AS ranking 
	FROM
		a
		LEFT JOIN dwd.dim_erp_sku_package_vol_info b ON a.SKU = b.erp_sku 
	) SELECT
	SPU,
	SKU,添加时间,产品分类,成本价,
	b.erp_package_vol 
FROM
	b 
WHERE
	ranking = 1 
	AND month_diff IS NULL
"""
    df = pd.read_sql(sql, engine)

得到每个SKU的最长边，围长，总重量，6000抛重，采购体积比（采购/6000抛重）

In [ ]:

import json
import re

import numpy as np

from sell.sell_price import call_sell_and_order_price
def extract_number(value):
    # 提取字符串中的第一个数字
        match = re.search(r"[-+]?\d*\.\d+|\d+", str(value))
        return float(match.group()) if match else 0.0
for index, row in df.iterrows():
    package_dict = json.loads(row['erp_package_vol'])
    max_length = 0 # 最长边
    max_girth = 0 # 最大围长
    all_weight = 0 # 总重量
    all_vol_weight = 0 # 总抛重
    for key, package in package_dict.items():
        package['长'] = extract_number(package['长'])
        package['宽'] = extract_number(package['宽'])
        package['高'] = extract_number(package['高'])
        package['重量'] = extract_number(package['重量'])
        size =sorted([package['长'],package['宽'],package['高']])
        fst_size = size[2]
        snd_size = size[1]
        thd_size = size[0]
        max_length=max(max_length,fst_size)
        max_girth=max(max_girth,fst_size+(snd_size+thd_size)*2)
        all_weight+=package['重量']/1000
        all_vol_weight+=package['长']*package['宽']*package['高']/6000
    sell_price,order_price,order_type = call_sell_and_order_price(row['成本价'], row['erp_package_vol'],"海运")
    df.loc[index,'网站售价'] = sell_price[0]
    df.loc[index,'物流分摊费'] = sell_price[1]
    df.loc[index,'订单物流费'] = order_price
    df.loc[index,'尾端类型'] = order_type
    df.loc[index,'最长边'] = max_length
    df.loc[index,'最大围长'] = max_girth
    df.loc[index,'总重量'] = all_weight
    df.loc[index,'总抛重'] = all_vol_weight
    print(index)

In [ ]:

# 按照那个分组，按照总抛重，每5总抛重为一组，最长边取大，最短边取小，最大实重取大，最小实重取小，网站售价求和，物流分摊费求和，订单物流费求和，尾端类型不要，
cost_bins = list(range(0, 4000, 10)) +[28700]
df['成本价分组'] = pd.cut(df['成本价'], bins=cost_bins, right=True, labels=cost_bins[1:])

# 2. 总抛重分组（按5为一组，0-5 为一组，5.01-10 为一组，等）
df['总抛重分组'] = (np.ceil(df['总抛重'] / 5) * 5).astype(int)
df = df.dropna(subset=['成本价分组'])
# 3. 分组聚合
agg_df = df.groupby(['成本价分组', '总抛重分组'], observed=True).agg({
    '最长边': ['max', 'min'],      # 每组最大 每组最小
    '最大围长': 'max',
    '总重量': ['max', 'min','sum'],  # 分别取最大/最小实重
    '网站售价': 'sum',
    '物流分摊费': 'sum',
    '订单物流费': 'sum',
    'SKU': 'count'
}).reset_index()

# 4. 重命名列
agg_df.columns = [
    '成本价分组', '总抛重分组',
    '最长边max', '最长边min', '最大围长',
    '总重量max', '总重量min','总重量',
    '网站售价', '物流分摊费', '订单物流费','SKU种类'
]
agg_df.to_clipboard(index=False)