config.py 22 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292
  1. class CustConfig:
  2. FEATURE_COLUMNS = [
  3. "BB_RETAIL_CUSTOMER_CODE", # 零售户代码
  4. "BB_RTL_CUST_GRADE_NAME", # 零售户分档名称
  5. "BB_RTL_CUST_MARKET_TYPE_NAME", # 零售户市场类型名称
  6. "STORE_AREA", # 店铺经营面积
  7. "BB_RTL_CUST_BUSINESS_TYPE_NAME", # 零售户业态名称
  8. "OPERATOR_EDU_LEVEL", # 零售客户经营者文化程
  9. "OPERATOR_AGE", # 经营者年龄
  10. "BB_RTL_CUST_CHAIN_FLAG", # 零售户连锁标识
  11. "PRESENT_STAR_TERMINAL", # 终端星级
  12. "MD04_MG_RTL_CUST_CREDITCLASS_NAME", # 零售户信用等级名称
  13. "MD04_DIR_SAL_STORE_FLAG", # 直营店标识
  14. "BB_CUSTOMER_MANAGER_SCOPE_NAME", # 零售户经营范围名称
  15. "PRODUCT_INSALE_QTY", # 在销品规数
  16. # "CUST_INVESTMENT", # 店铺资源投入建设
  17. # "NEW_PRODUCT_MEMBERS_QTY_SAMEPRICE_OCC", # 新品订货量占同价类比重
  18. # "PRODUCT_LISTING_RATE", # 品规上架率
  19. # "STOCKOUT_DAYS", # 断货天数
  20. # "YLT_TURNOVER_RATE", # 易灵通动销率
  21. # "YLT_BAR_PACKAGE_SALE_OCC", # 易灵通条包销售占比
  22. # "UNPACKING_RATE", # 拆包率
  23. # "BB_RTL_CUST_POSITION_TYPE_NAME", # 零售户商圈类型名称
  24. # "BB_RTL_CUST_SUB_BUSI_PLACE_NAME", # 零售户业态细分名称
  25. # "BB_RTL_CUST_TERMINAL_LEVEL_NAME", # 零售户终端层级名称
  26. # "BB_RTL_CUST_TERMINALEVEL_NAME", # 零售户终端层级细分名称
  27. # "MD04_MG_SAMPLE_CUST_FLAG", # 样本户标识
  28. # "MD07_RTL_CUST_IS_SALE_LARGE_FLAG", # 零售户大户标识
  29. # "BB_RTL_CUST_OPERATE_METHOD_NAME", # 零售户经营方式名称
  30. # "BB_RTL_CUST_CGT_OPERATE_SCOPE_NAME", # 零售户卷烟经营规模名称
  31. # "AVERAGE_CONSUMER_FLOW", # 月均消费人流
  32. # "NEW_PRODUCT_MEMBERS_QTY", # 新品消费会员数量
  33. ]
  34. # 数据清洗规则
  35. CLEANING_RULES = {
  36. "BB_RTL_CUST_GRADE_NAME": {"method": "fillna", "opt": "fill", "value": "十五档", "type": "str"},
  37. "BB_RTL_CUST_MARKET_TYPE_NAME": {"method": "fillna", "opt": "fill", "value": "城网", "type": "str"},
  38. "STORE_AREA": {"method": "fillna", "opt": "mean", "type": "num"},
  39. "BB_RTL_CUST_BUSINESS_TYPE_NAME": {"method": "fillna", "opt": "fill", "value": "其他", "type": "str"},
  40. "OPERATOR_EDU_LEVEL": {"method": "fillna", "opt": "fill", "value": "无数据", "type": "str"},
  41. "OPERATOR_AGE": {"method": "fillna", "opt": "mean", "type": "num"},
  42. "BB_RTL_CUST_CHAIN_FLAG": {"method": "fillna", "opt": "fill", "value": "否", "type": "str"},
  43. "PRESENT_STAR_TERMINAL": {"method": "fillna", "opt": "fill", "value": "非星级", "type": "str"},
  44. "MD04_MG_RTL_CUST_CREDITCLASS_NAME": {"method": "fillna", "opt": "fill", "value": "B", "type": "str"},
  45. "MD04_DIR_SAL_STORE_FLAG": {"method": "fillna", "opt": "fill", "value": "否", "type": "str"},
  46. "BB_CUSTOMER_MANAGER_SCOPE_NAME": {"method": "fillna", "opt": "fill", "value": "否", "type": "str"},
  47. "PRODUCT_INSALE_QTY": {"method": "fillna", "opt": "mean", "type": "num"},
  48. # "CUST_INVESTMENT": {"method": "fillna", "opt": "fill", "type": 0}
  49. # "NEW_PRODUCT_MEMBERS_QTY_SAMEPRICE_OCC": {"method": "fillna", "opt": "mean", "type": "num"},
  50. # "PRODUCT_LISTING_RATE": {"method": "fillna", "opt": "mean", "type": "num"},
  51. # "STOCKOUT_DAYS": {"method": "fillna", "opt": "mean", "type": "num"},
  52. # "YLT_TURNOVER_RATE": {"method": "fillna", "opt": "mean", "type": "num"},
  53. # "NEW_PRODUCT_MEMBERS_QTY": {"method": "fillna", "opt": "mean", "type": "num"},
  54. # "PRODUCT_INSALE_QTY": {"method": "fillna", "opt": "mean", "type": "num"},
  55. # "UNPACKING_RATE": {"method": "fillna", "opt": "mean", "type": "num"},
  56. # "BB_RTL_CUST_POSITION_TYPE_NAME": {"method": "fillna", "opt": "fill", "value": "其他", "type": "str"},
  57. # "BB_RTL_CUST_SUB_BUSI_PLACE_NAME": {"method": "fillna", "opt": "fill", "value": "其他", "type": "str"},
  58. # "BB_RTL_CUST_TERMINALEVEL_NAME": {"method": "fillna", "opt": "replace", "value": "BB_RTL_CUST_TERMINAL_LEVEL_NAME", "type": "str"},
  59. # "MD04_MG_SAMPLE_CUST_FLAG": {"method": "fillna", "value": "N", "opt": "fill"},
  60. # "MD07_RTL_CUST_IS_SALE_LARGE_FLAG": {"method": "fillna", "value": "N", "opt": "fill"},
  61. # "BB_RTL_CUST_CGT_OPERATE_SCOPE_NAME": {"method": "fillna", "value": "中", "opt": "fill"},
  62. }
  63. ONEHOT_CAT = {
  64. "BB_RTL_CUST_GRADE_NAME": ['一档', '二档', '三档', '四档', '五档', '六档', '七档', '八档', '九档', '十档', '十一档', '十二档',
  65. '十三档', '十四档', '十五档', '十六档', '十七档', '十八档', '十九档', '二十档', '二十一档', '二十二档',
  66. '二十三档', '二十四档', '二十五档', '二十六档', '二十七档', '二十八档', '二十九档', '三十档'],
  67. "BB_RTL_CUST_MARKET_TYPE_NAME": ["城网", "农网"],
  68. "BB_RTL_CUST_BUSINESS_TYPE_NAME": ["便利店", "超市", "烟草专业店", "娱乐服务类", "其他"],
  69. "OPERATOR_EDU_LEVEL": [1, 2, 3, 4, 5, 6, 7, "无数据"],
  70. "BB_RTL_CUST_CHAIN_FLAG": ["是", "否"],
  71. "PRESENT_STAR_TERMINAL": ["一星", "二星", "三星", "四星", "五星", "非星级"],
  72. "MD04_MG_RTL_CUST_CREDITCLASS_NAME": ["AAA", "AA", "A", "B", "C", "D"],
  73. "MD04_DIR_SAL_STORE_FLAG": ["是", "否"],
  74. "BB_CUSTOMER_MANAGER_SCOPE_NAME": ["是", "否"],
  75. # "BB_RTL_CUST_POSITION_TYPE_NAME": ["居民区", "商业娱乐区", "交通枢纽区", "旅游景区", "工业区", "集贸区", "院校学区", "办公区", "其他"]
  76. }
  77. class ProductConfig:
  78. FEATURE_COLUMNS = [
  79. "product_code", # 商品编码
  80. "direct_retail_price", # 建议零售价
  81. "is_low_tar", # 是否低焦油烟
  82. "tar_qty", # 焦油含量
  83. "is_exploding_beads", # 是否爆珠
  84. "is_shortbranch", # 是否短支烟
  85. "is_medium", # 是否中支烟
  86. "is_tiny", # 是否细支
  87. "product_style_code_name", # 包装类型名称
  88. "org_is_abnormity", # 是否异形包装
  89. "is_chuangxin", # 是否创新品类
  90. "is_key_brand", # 是否重点品牌
  91. "foster_level_hy", # 是否行业共育品规
  92. "foster_level_sj", # 是否省级共育品规
  93. "is_cigar", # 是否雪茄型卷烟
  94. "co_qty", # 一氧化碳含量
  95. "tbc_total_length", # 烟支总长度
  96. "tbc_length", # 烟支长度
  97. "filter_length", # 滤嘴长度
  98. # "adjust_price", # 含税调拨价
  99. # "notwithtax_adjust_price", # 不含税调拨价
  100. # "whole_sale_price", # 统一批发价
  101. # "allot_price", # 调拨价
  102. # "direct_whole_price", # 批发指导价
  103. # "retail_price", # 零售价
  104. # "price_type_name", # 卷烟价类名称
  105. # "gear_type_name", # 卷烟档位名称
  106. # "category_type_name", # 卷烟品类名称
  107. # "is_high_level", # 是否高端烟
  108. # "is_upscale_level", # 是否高端烟不含高价
  109. # "is_high_price", # 是否高价烟
  110. # "is_low_price", # 是否低价烟
  111. # "is_encourage", # 是否全国鼓励品牌
  112. # "is_abnormity", # 是否异形包装
  113. # "is_intake", # 是否进口烟
  114. # "is_short", # 是否紧俏品牌
  115. # "is_ordinary_price_type", # 是否普一类烟
  116. # "source_type", # 来源类型
  117. # "chinese_mix", # 中式混合
  118. # "sub_price_type_name", # 细分卷烟价类名称
  119. ]
  120. CLEANING_RULES = {
  121. "direct_retail_price": {"method": "fillna", "opt": "mean", "type": "num"},
  122. "is_low_tar": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  123. "tar_qty": {"method": "fillna", "opt": "mean", "type": "num"},
  124. "is_exploding_beads": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  125. "is_shortbranch": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  126. "is_medium": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  127. "is_tiny": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  128. "product_style_code_name": {"method": "fillna", "opt": "fill", "type": "str", "value": "其他"},
  129. "org_is_abnormity": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  130. "is_chuangxin": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  131. "is_key_brand": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  132. "foster_level_hy": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  133. "foster_level_sj": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  134. "is_cigar": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  135. "co_qty": {"method": "fillna", "opt": "mean", "type": "num"},
  136. "tbc_total_length": {"method": "fillna", "opt": "mean", "type": "num"},
  137. "tbc_length": {"method": "fillna", "opt": "mean", "type": "num"},
  138. "filter_length": {"method": "fillna", "opt": "mean", "type": "num"},
  139. # "adjust_price": {"method": "fillna", "opt": "mean", "type": "num"},
  140. # "notwithtax_adjust_price": {"method": "fillna", "opt": "mean", "type": "num"},
  141. # "whole_sale_price": {"method": "fillna", "opt": "mean", "type": "num"},
  142. # "allot_price": {"method": "fillna", "opt": "fill", "type": "num", "value": 0.0},
  143. # "direct_whole_price": {"method": "fillna", "opt": "mean", "type": "num"},
  144. # "retail_price": {"method": "fillna", "opt": "mean", "type": "num"},
  145. # "price_type_name": {"method": "fillna", "opt": "fill", "type": "str", "value": "一类烟"},
  146. # "gear_type_name": {"method": "fillna", "opt": "fill", "type": "str", "value": "其他"},
  147. # "category_type_name": {"method": "fillna", "opt": "fill", "type": "str", "value": "其他"},
  148. # "is_high_level": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  149. # "is_upscale_level": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  150. # "is_high_price": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  151. # "is_low_price": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  152. # "is_encourage": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  153. # "is_abnormity": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  154. # "is_intake": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  155. # "is_short": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  156. # "is_ordinary_price_type": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  157. # "source_type": {"method": "fillna", "opt": "fill", "type": "str", "value": "其他"},
  158. # "chinese_mix": {"method": "fillna", "opt": "fill", "type": "str", "value": "否"},
  159. # "sub_price_type_name": {"method": "fillna", "opt": "fill", "type": "str", "value": "普一类烟"},
  160. }
  161. ONEHOT_CAT = {
  162. "is_low_tar": ["是", "否"],
  163. "is_exploding_beads": ["是", "否"],
  164. "is_shortbranch": ["是", "否"],
  165. "is_medium": ["是", "否"],
  166. "is_tiny": ["是", "否"],
  167. "product_style_code_name": ["条盒硬盒", "条包硬盒", "条盒软盒", "条包软盒", "铁盒", "其他"],
  168. "org_is_abnormity": ["是", "否"],
  169. "is_chuangxin": ["是", "否"],
  170. "is_key_brand": ["是", "否"],
  171. "foster_level_hy": ["是", "否"],
  172. "foster_level_sj": ["是", "否"],
  173. "is_cigar": ["是", "否"],
  174. # "price_type_name": ["一类烟", "二类烟", "三类烟", "四类烟", "五类烟", "无价类"],
  175. # "gear_type_name": ["第一档位", "第二档位", "第三档位", "第四档位", "第五档位", "第六档位", "第七档位", "第八档位", "其他"],
  176. # "category_type_name": ["第1品类", "第2品类", "第3品类", "第4品类", "第5品类", "第6品类", "第7品类",
  177. # "第8品类", "第9品类", "第10品类", "第11品类", "第12品类", "第13品类", "其他"],
  178. # "is_high_level": ["是", "否"],
  179. # "is_upscale_level": ["是", "否"],
  180. # "is_high_price": ["是", "否"],
  181. # "is_low_price": ["是", "否"],
  182. # "is_encourage": ["是", "否"],
  183. # "is_abnormity": ["是", "否"],
  184. # "is_intake": ["是", "否"],
  185. # "is_short": ["是", "否"],
  186. # "is_ordinary_price_type": ["是", "否"],
  187. # "source_type": ["是", "否"],
  188. # "chinese_mix": ["是", "否"],
  189. # "sub_price_type_name": ["高端烟", "高价位烟", "普一类烟", "二类烟", "三类烟", "四类烟", "五类烟", "无价类"],
  190. }
  191. class OrderConfig:
  192. FEATURE_COLUMNS = [
  193. "BB_RETAIL_CUSTOMER_CODE", # 零售户编码
  194. "PRODUCT_CODE", # 卷烟编码
  195. "MONTH6_SALE_QTY", # 近半年销量(箱)
  196. "MONTH6_SALE_AMT", # 近半年销售额(万元)
  197. "MONTH6_GROSS_PROFIT_RATE", # 近半年毛利率
  198. "MONTH6_SALE_QTY_YOY", # 销售量同比
  199. "MONTH6_SALE_QTY_MOM", # 销售量环比
  200. "MONTH6_SALE_AMT_YOY", # 销售额(购进额)同比
  201. "MONTH6_SALE_AMT_MOM", # 销售额(购进额)环比
  202. "STOCK_QTY", # 库存
  203. "ORDER_FULLORDR_RATE", # 订足率
  204. "FULL_FILLMENT_RATE", # 订单满足率
  205. "ORDER_FULLORDR_RATE_MOM", # 订足率环比
  206. "CUSTOMER_REPURCHASE_RATE", # 会员重购率
  207. "DEMAND_RATE", # 需求量满足率
  208. "LISTING_RATE", # 品规商上架率
  209. "PUT_MARKET_FINISH_RATE", # 投放完成率
  210. "OUT_STOCK_DAYS", # 断货天数
  211. "YLT_TURNOVER_RATE", # 易灵通动销率
  212. "YLT_BAR_PACKAGE_SALE_OCC", # 易灵通调包销售占比
  213. "UNPACKING_RATE", # 拆包率
  214. "POS_PACKAGE_PRICE", # pos机单包价格
  215. ]
  216. WEIGHTS = {
  217. "MONTH6_SALE_QTY": 0.15,
  218. "MONTH6_SALE_QTY_MOM": 0.2,
  219. "ORDER_FULLORDR_RATE": 0.3,
  220. "ORDER_FULLORDR_RATE_MOM": 0.35,
  221. }
  222. class ImportanceFeaturesMap:
  223. CUSTOM_FEATRUES_MAP = {
  224. "BB_RTL_CUST_GRADE_NAME": "零售户分档名称",
  225. "BB_RTL_CUST_MARKET_TYPE_NAME": "零售户市场类型名称",
  226. "STORE_AREA": "店铺经营面积",
  227. "BB_RTL_CUST_BUSINESS_TYPE_NAME": "零售户业态名称",
  228. "OPERATOR_EDU_LEVEL": "零售客户经营者文化程",
  229. "OPERATOR_AGE": "经营者年龄",
  230. "BB_RTL_CUST_CHAIN_FLAG": "零售户连锁标识",
  231. "PRESENT_STAR_TERMINAL": "终端星级",
  232. "MD04_MG_RTL_CUST_CREDITCLASS_NAME": "零售户信用等级名称",
  233. "MD04_DIR_SAL_STORE_FLAG": "直营店标识",
  234. "BB_CUSTOMER_MANAGER_SCOPE_NAME": "零售户经营范围名称",
  235. "PRODUCT_INSALE_QTY": "在销品规数",
  236. # "CUST_INVESTMENT": "店铺资源投入建设",
  237. }
  238. PRODUCT_FEATRUES_MAP = {
  239. # ProductConfig 字段映射
  240. "direct_retail_price": "建议零售价",
  241. "is_low_tar": "是否低焦油烟",
  242. "tar_qty": "焦油含量",
  243. "is_exploding_beads": "是否爆珠",
  244. "is_shortbranch": "是否短支烟",
  245. "is_medium": "是否中支烟",
  246. "is_tiny": "是否细支",
  247. "product_style_code_name": "包装类型名称",
  248. "org_is_abnormity": "是否异形包装",
  249. "is_chuangxin": "是否创新品类",
  250. "is_key_brand": "是否重点品牌",
  251. "foster_level_hy": "是否行业共育品规",
  252. "foster_level_sj": "是否省级共育品规",
  253. "is_cigar": "是否雪茄型卷烟",
  254. "co_qty": "一氧化碳含量",
  255. "tbc_total_length": "烟支总长度",
  256. "tbc_length": "烟支长度",
  257. "filter_length": "滤嘴长度",
  258. }