constant.py 7.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329
  1. general_columns = [
  2. ]
  3. general_drop_columns = [
  4. "THEMONTH_V",
  5. "EVENT_CPNAME_C",
  6. "EVENT_SPNAME_C",
  7. "EVENT_CATEGORYNAME_C",
  8. "EVENT_CHANNEL_C",
  9. "EVENT_CHANNEL_BELONGTO_C",
  10. "EVENT_ORDER_MONTH_C",
  11. "EVENT_ORDER_MONTH_V",
  12. "EVENT_CANCEL_DIFF_C",
  13. "EVENT_CANCEL_DIFF_V",
  14. "EVENT_ORDER_PRICE_C",
  15. "EVENT_ORDER_PRICE_V",
  16. "EVENT_ORDER_FLOWP_C",
  17. "EVENT_ORDER_FLOWP_V",
  18. "EVENT_ORDER_STATE_C",
  19. "EVENT_ORDER_SUMMONTH_C",
  20. "EVENT_ORDER_SUMMONTH_V",
  21. "TAG_INTIME_V",
  22. "TAG_PACKAGE_C",
  23. "TAG_PACKAGE_PIC_C",
  24. "EVENT_PHONETYPE_C",
  25. "TAG_AREA_C",
  26. "EVENT_APP_USE.V",
  27. "I_appuse"
  28. ]
  29. origin_column_names = [
  30. "THEMONTH_V",
  31. "uuid",
  32. "TAG_AREA_C",
  33. "EVENT_CPNAME_C",
  34. "EVENT_SPNAME_C",
  35. "EVENT_CATEGORYNAME_C",
  36. "EVENT_CHANNEL_C",
  37. "EVENT_CHANNEL_BELONGTO_C",
  38. "EVENT_ORDER_MONTH_C",
  39. "EVENT_ORDER_MONTH_V",
  40. "EVENT_CANCEL_DIFF_C",
  41. "EVENT_CANCEL_DIFF_V",
  42. "EVENT_ORDER_PRICE_C",
  43. "EVENT_ORDER_PRICE_V",
  44. "EVENT_ORDER_FLOWP_C",
  45. "EVENT_ORDER_FLOWP_V",
  46. "EVENT_ORDER_STATE_C",
  47. "EVENT_ORDER_SUMMONTH_C",
  48. "EVENT_ORDER_SUMMONTH_V",
  49. "TAG_INTIME_C",
  50. "TAG_INTIME_V",
  51. "TAG_PACKAGE_C",
  52. "TAG_PACKAGE_PIC_C",
  53. "EVENT_FLUX_C",
  54. "EVENT_FLUX_V",
  55. "EVENT_CONSUM_C",
  56. "EVENT_CONSUM_V",
  57. "EVENT_PHONETYPE_C",
  58. "EVENT_VIDEO_FLUX_C",
  59. "EVENT_VIDEO_FLUX_V",
  60. "EVENT_IS_ACCT_C",
  61. "TAG_GENDER_C",
  62. "TAG_PROVINCE_C",
  63. "TAG_NETTYPE_C",
  64. "TAG_AGE_C",
  65. "EVENT_APP_USE.C",
  66. "EVENT_APP_USE.V",
  67. "EVENT_USER_OSTATE_C",
  68. "I_appuse"
  69. ]
  70. low_dim_cols = [
  71. "uuid",
  72. "MAvg_TOTAL_FLUX_1_3_zs",
  73. "MPer1_TOTAL_FLUX_zs",
  74. "MAvg_TOTAL_VIDEO_FLUX_1_3_zs",
  75. "MPer1_TOTAL_VIDEO_FLUX_zs",
  76. "MAvg_Flow_kuaishou_1_3_zs",
  77. "MPer1_Flow_kuaishou_zs",
  78. "Div_kuaishou_vFlux_1_3"
  79. ]
  80. pkl_columns = [
  81. "row_id",
  82. "month",
  83. "EVENT_FLUX_V",
  84. "EVENT_CONSUM_V",
  85. "EVENT_VIDEO_FLUX_V",
  86. "TAG_GENDER_C_0",
  87. "TAG_GENDER_C_1",
  88. "TAG_GENDER_C_2",
  89. "TAG_NETTYPE_C_0",
  90. "TAG_NETTYPE_C_1",
  91. "TAG_NETTYPE_C_2",
  92. "TAG_NETTYPE_C_3",
  93. "TAG_NETTYPE_C_4",
  94. "TAG_AGE_C_0",
  95. "TAG_AGE_C_1",
  96. "TAG_AGE_C_2",
  97. "TAG_AGE_C_3",
  98. "TAG_AGE_C_4",
  99. "TAG_AGE_C_5",
  100. "TAG_INTIME_C_0",
  101. "TAG_INTIME_C_1",
  102. "TAG_INTIME_C_2",
  103. "TAG_INTIME_C_3",
  104. "TAG_INTIME_C_4",
  105. "TAG_INTIME_C_5",
  106. "EVENT_IS_ACCT_C_0",
  107. "EVENT_IS_ACCT_C_1",
  108. "EVENT_IS_ACCT_C_2",
  109. "EVENT_FLUX_C_0",
  110. "EVENT_FLUX_C_1",
  111. "EVENT_FLUX_C_2",
  112. "EVENT_FLUX_C_3",
  113. "EVENT_FLUX_C_4",
  114. "EVENT_FLUX_C_5",
  115. "EVENT_FLUX_C_6",
  116. "EVENT_FLUX_C_7",
  117. "EVENT_FLUX_C_8",
  118. "EVENT_CONSUM_C_0",
  119. "EVENT_CONSUM_C_1",
  120. "EVENT_CONSUM_C_2",
  121. "EVENT_CONSUM_C_3",
  122. "EVENT_CONSUM_C_4",
  123. "EVENT_CONSUM_C_5",
  124. "EVENT_CONSUM_C_6",
  125. "EVENT_CONSUM_C_7",
  126. "EVENT_VIDEO_FLUX_C_0",
  127. "EVENT_VIDEO_FLUX_C_1",
  128. "EVENT_VIDEO_FLUX_C_2",
  129. "EVENT_VIDEO_FLUX_C_3",
  130. "EVENT_VIDEO_FLUX_C_4",
  131. "EVENT_VIDEO_FLUX_C_5",
  132. "EVENT_VIDEO_FLUX_C_6",
  133. "EVENT_VIDEO_FLUX_C_7",
  134. "TAG_PROVINCE_C_0",
  135. "TAG_PROVINCE_C_1",
  136. "TAG_PROVINCE_C_2",
  137. "TAG_PROVINCE_C_3",
  138. "TAG_PROVINCE_C_4",
  139. "TAG_PROVINCE_C_5",
  140. "TAG_PROVINCE_C_6",
  141. "TAG_PROVINCE_C_7",
  142. "TAG_PROVINCE_C_8",
  143. "TAG_PROVINCE_C_9",
  144. "TAG_PROVINCE_C_10",
  145. "TAG_PROVINCE_C_11",
  146. "TAG_PROVINCE_C_12",
  147. "TAG_PROVINCE_C_13",
  148. "TAG_PROVINCE_C_14",
  149. "TAG_PROVINCE_C_15",
  150. "TAG_PROVINCE_C_16",
  151. "TAG_PROVINCE_C_17",
  152. "TAG_PROVINCE_C_18",
  153. "TAG_PROVINCE_C_19",
  154. "TAG_PROVINCE_C_20",
  155. "TAG_PROVINCE_C_21",
  156. "TAG_PROVINCE_C_22",
  157. "TAG_PROVINCE_C_23",
  158. "TAG_PROVINCE_C_24",
  159. "TAG_PROVINCE_C_25",
  160. "TAG_PROVINCE_C_26",
  161. "TAG_PROVINCE_C_27",
  162. "TAG_PROVINCE_C_28",
  163. "TAG_PROVINCE_C_29",
  164. "TAG_PROVINCE_C_30",
  165. "TAG_PROVINCE_C_31",
  166. "app_use_tencent_0",
  167. "app_use_tencent_1",
  168. "app_use_tencent_2",
  169. "app_use_tencent_3",
  170. "app_use_tencent_4",
  171. "app_use_tencent_5",
  172. "app_use_tencent_6",
  173. "app_use_tencent_7",
  174. "app_use_mangguo_0",
  175. "app_use_mangguo_1",
  176. "app_use_mangguo_2",
  177. "app_use_mangguo_3",
  178. "app_use_mangguo_4",
  179. "app_use_mangguo_5",
  180. "app_use_mangguo_6",
  181. "app_use_mangguo_7",
  182. "app_use_youku_0",
  183. "app_use_youku_1",
  184. "app_use_youku_2",
  185. "app_use_youku_3",
  186. "app_use_youku_4",
  187. "app_use_youku_5",
  188. "app_use_youku_6",
  189. "app_use_youku_7",
  190. "app_use_iqiyi_0",
  191. "app_use_iqiyi_1",
  192. "app_use_iqiyi_2",
  193. "app_use_iqiyi_3",
  194. "app_use_iqiyi_4",
  195. "app_use_iqiyi_5",
  196. "app_use_iqiyi_6",
  197. "app_use_iqiyi_7",
  198. "app_use_bilibili_0",
  199. "app_use_bilibili_1",
  200. "app_use_bilibili_2",
  201. "app_use_bilibili_3",
  202. "app_use_bilibili_4",
  203. "app_use_bilibili_5",
  204. "app_use_bilibili_6",
  205. "app_use_bilibili_7",
  206. "app_use_kuaishou_0",
  207. "app_use_kuaishou_1",
  208. "app_use_kuaishou_2",
  209. "app_use_kuaishou_3",
  210. "app_use_kuaishou_4",
  211. "app_use_kuaishou_5",
  212. "app_use_kuaishou_6",
  213. "app_use_kuaishou_7"
  214. ]
  215. process_column_names = [
  216. "row_id",
  217. "month",
  218. "EVENT_CATEGORYNAME_C",
  219. "EVENT_CHANNEL_BELONGTO_C",
  220. "EVENT_CANCEL_DIFF_C",
  221. "TAG_INTIME_C",
  222. "EVENT_FLUX_C",
  223. "EVENT_FLUX_V",
  224. "EVENT_CONSUM_C",
  225. "EVENT_CONSUM_V",
  226. "EVENT_VIDEO_FLUX_C",
  227. "EVENT_VIDEO_FLUX_V",
  228. "TAG_GENDER_C",
  229. "TAG_NETTYPE_C",
  230. "TAG_AGE_C",
  231. "app_use_tencent",
  232. "app_use_mangguo",
  233. "app_use_youku",
  234. "app_use_iqiyi",
  235. "app_use_bilibili",
  236. "app_use_kuaishou"
  237. ]
  238. feature_column_names = [
  239. "row_id",
  240. "month",
  241. "TAG_INTIME_C_0",
  242. "TAG_INTIME_C_1",
  243. "TAG_INTIME_C_2",
  244. "TAG_INTIME_C_3",
  245. "TAG_INTIME_C_4",
  246. "TAG_INTIME_C_5",
  247. "EVENT_FLUX_C",
  248. "EVENT_FLUX_V",
  249. "EVENT_CONSUM_V",
  250. "EVENT_VIDEO_FLUX_V",
  251. "EVENT_CONSUM_C",
  252. "EVENT_VIDEO_FLUX_C",
  253. "app_use_tencent",
  254. "app_use_mangguo",
  255. "app_use_youku",
  256. "app_use_iqiyi",
  257. "app_use_bilibili",
  258. "app_use_kuaishou",
  259. "TAG_GENDER_C_0",
  260. "TAG_GENDER_C_1",
  261. "TAG_GENDER_C_2",
  262. "TAG_NETTYPE_C_1",
  263. "TAG_NETTYPE_C_2",
  264. "TAG_NETTYPE_C_3",
  265. "TAG_NETTYPE_C_4",
  266. "TAG_NETTYPE_C_5",
  267. "EVENT_CATEGORYNAME_C_0",
  268. "EVENT_CATEGORYNAME_C_1",
  269. "EVENT_CATEGORYNAME_C_2",
  270. "EVENT_CATEGORYNAME_C_3",
  271. "EVENT_CATEGORYNAME_C_4",
  272. "EVENT_CATEGORYNAME_C_5",
  273. "EVENT_CATEGORYNAME_C_6",
  274. "EVENT_CATEGORYNAME_C_7",
  275. "EVENT_CATEGORYNAME_C_8",
  276. "EVENT_CATEGORYNAME_C_9",
  277. "EVENT_CATEGORYNAME_C_10",
  278. "EVENT_CHANNEL_BELONGTO_C_0",
  279. "EVENT_CHANNEL_BELONGTO_C_1",
  280. "EVENT_CHANNEL_BELONGTO_C_2",
  281. "EVENT_CHANNEL_BELONGTO_C_3",
  282. "EVENT_CHANNEL_BELONGTO_C_4",
  283. "EVENT_CHANNEL_BELONGTO_C_5",
  284. "EVENT_CHANNEL_BELONGTO_C_6",
  285. "TAG_AGE_C_0",
  286. "TAG_AGE_C_1",
  287. "TAG_AGE_C_2",
  288. "TAG_AGE_C_3",
  289. "TAG_AGE_C_4",
  290. "TAG_AGE_C_5",
  291. "EVENT_CANCEL_DIFF_C_0",
  292. "EVENT_CANCEL_DIFF_C_1",
  293. "EVENT_CANCEL_DIFF_C_2",
  294. "EVENT_CANCEL_DIFF_C_3",
  295. "EVENT_CANCEL_DIFF_C_4",
  296. "EVENT_CANCEL_DIFF_C_5",
  297. "EVENT_CANCEL_DIFF_C_6"
  298. ]
  299. insert_origin_tb_name = "lu_origin_data_neg"
  300. insert_pos_tb_name = "lu_pos_data_0218_07"
  301. insert_pos_tb_test_name = "lu_pos_data_test_7"
  302. insert_neg_tb_test_name = "lu_neg_data_test_7"
  303. insert_neg_tb_name = "lu_neg_data_0218_07"
  304. def is_number(s):
  305. try:
  306. float(s)
  307. return True
  308. except ValueError:
  309. pass
  310. try:
  311. import unicodedata
  312. unicodedata.numeric(s)
  313. return True
  314. except (TypeError, ValueError):
  315. pass
  316. return False