{"id":9914,"date":"2018-04-24T19:21:50","date_gmt":"2018-04-24T11:21:50","guid":{"rendered":"https:\/\/www.growthhk.cn\/?p=9914"},"modified":"2018-04-24T19:22:25","modified_gmt":"2018-04-24T11:22:25","slug":"9914","status":"publish","type":"post","link":"https:\/\/www.growthhk.cn\/cgo\/9914.html","title":{"rendered":"CGO\u53c2\u8003\uff1a\u643a\u7a0b\u4e2a\u6027\u5316\u63a8\u8350\u7b97\u6cd5\u5b9e\u8df5"},"content":{"rendered":"<p>\u643a\u7a0b\u57fa\u7840\u4e1a\u52a1\u7814\u53d1\u90e8-\u6570\u636e\u4ea7\u54c1\u548c\u670d\u52a1\u7ec4\uff0c\u4e13\u6ce8\u4e8e<span class=\"wpcom_tag_link\"><a href=\"https:\/\/www.growthhk.cn\/tag\/%e4%b8%aa%e6%80%a7%e5%8c%96%e6%8e%a8%e8%8d%90\" title=\"\u4e2a\u6027\u5316\u63a8\u8350\" target=\"_blank\">\u4e2a\u6027\u5316\u63a8\u8350<\/a><\/span>\u3001\u81ea\u7136\u8bed\u8a00\u5904\u7406\u3001\u56fe\u50cf\u8bc6\u522b\u7b49\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u5148\u8fdb\u6280\u672f\u5728\u65c5\u6e38\u884c\u4e1a\u7684\u5e94\u7528\u7814\u7a76\u5e76\u843d\u5730\u4ea7\u751f\u4ef7\u503c\u3002<\/p>\n<p>\u76ee\u524d\uff0c\u56e2\u961f\u5df2\u7ecf\u4e3a\u643a\u7a0b\u63d0\u4f9b\u4e86\u901a\u7528\u5316\u7684\u4e2a\u6027\u5316\u63a8\u8350\u7cfb\u7edf\u3001\u667a\u80fd\u5ba2\u670d\u7cfb\u7edf\u3001AI\u5e73\u53f0\u7b49\u4e00\u7cfb\u5217\u6210\u719f\u7684\u4ea7\u54c1\u4e0e\u670d\u52a1\u3002<\/p>\n<p>\u643a\u7a0b\u4f5c\u4e3a\u56fd\u5185\u9886\u5148\u7684 OTA\uff0c\u6bcf\u5929\u5411\u4e0a\u5343\u4e07\u7528\u6237\u63d0\u4f9b\u5168\u65b9\u4f4d\u7684\u65c5\u884c\u670d\u52a1\uff0c\u5982\u4f55\u4e3a\u5982\u6b64\u4f17\u591a\u7684\u7528\u6237\u53d1\u73b0\u9002\u5408\u81ea\u5df1\u7684\u65c5\u6e38\u4ea7\u54c1\u4e0e\u670d\u52a1\uff0c\u6316\u6398\u6f5c\u5728\u7684\u5174\u8da3\uff0c\u7f13\u89e3\u4fe1\u606f\u8fc7\u8f7d\uff0c\u4e2a\u6027\u5316\u63a8\u8350\u7cfb\u7edf\u4e0e\u7b97\u6cd5\u5728\u5176\u4e2d\u53d1\u6325\u7740\u4e0d\u53ef\u6216\u7f3a\u7684\u4f5c\u7528\u3002<\/p>\n<p>\u800c OTA \u7684\u4e2a\u6027\u5316\u63a8\u8350\u4e00\u76f4\u4e5f\u662f\u4e2a\u96be\u70b9\uff0c\u6ca1\u6709\u592a\u591a\u6210\u529f\u7ecf\u9a8c\u53ef\u4ee5\u501f\u9274\uff0c\u672c\u6587\u5206\u4eab\u4e86\u643a\u7a0b\u5728\u4e2a\u6027\u5316\u63a8\u8350\u5b9e\u8df5\u4e2d\u7684\u4e00\u4e9b\u5c1d\u8bd5\u4e0e\u6478\u7d22\u3002<\/p>\n<p>\u63a8\u8350\u6d41\u7a0b\u5927\u4f53\u4e0a\u53ef\u4ee5\u5206\u4e3a3\u4e2a\u90e8\u5206\uff0c\u53ec\u56de\u3001\u6392\u5e8f\u3001\u63a8\u8350\u7ed3\u679c\u751f\u6210\uff0c\u6574\u4f53\u7684\u67b6\u6784\u5982\u4e0b\u56fe\u6240\u793a\u3002<\/p>\n<p><img decoding=\"async\" class=\"alignnone size-full wp-image-9915\" src=\"https:\/\/www.growthhk.cn\/wp-content\/uploads\/mmexport1524568219004.png\" alt=\"\" width=\"748\" height=\"395\" \/><\/p>\n<p>\u53ec\u56de\u9636\u6bb5\uff0c\u4e3b\u8981\u662f\u5229\u7528\u6570\u636e\u5de5\u7a0b\u548c\u7b97\u6cd5\u7684\u65b9\u5f0f\uff0c\u4ece\u5343\u4e07\u7ea7\u7684\u4ea7\u54c1\u4e2d\u9501\u5b9a\u7279\u5b9a\u7684\u5019\u9009\u96c6\u5408\uff0c\u5b8c\u6210\u5bf9\u4ea7\u54c1\u7684\u521d\u6b65\u7b5b\u9009\uff0c\u5176\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u51b3\u5b9a\u4e86\u6392\u5e8f\u9636\u6bb5\u7684\u6548\u7387\u548c\u63a8\u8350\u7ed3\u679c\u7684\u4f18\u52a3\u3002<\/p>\n<p>\u4e1a\u5185\u6bd4\u8f83\u4f20\u7edf\u7684\u7b97\u6cd5\uff0c\u4e3b\u8981\u662fCF[1][2]\u3001\u57fa\u4e8e\u7edf\u8ba1\u7684 Contextual \u63a8\u8350\u548cLBS\uff0c\u4f46\u8fd1\u671f\u6765\u6df1\u5ea6\u5b66\u4e60\u88ab\u5e7f\u6cdb\u5f15\u5165\uff0c\u7b97\u6cd5\u6027\u53d6\u5f97\u8f83\u5927\u7684\u63d0\u5347\uff0c\u5982\uff1a2015\u5e74 Netflix \u548c Gravity R&amp;D Inc \u63d0\u51fa\u7684\u5229\u7528 RNN\u7684Session-based \u63a8\u8350[5]\uff0c2016\u5e74Recsys\u4e0a\u63d0\u51fa\u7684\u7ed3\u5408 CNN \u548c <span class=\"wpcom_keyword_link\"><a href=\"https:\/\/www.growthhk.cn\/tag\/PMF\" target=\"_blank\" title=\"PMF\">PMF<\/a><\/span> \u5e94\u7528\u4e8e Context-aware \u63a8\u8350[10]\uff0c2016\u5e74 Google \u63d0\u51fa\u7684\u5c06 DNN \u4f5c\u4e3a MF \u7684\u63a8\u5e7f\uff0c\u53ef\u4ee5\u5f88\u5bb9\u6613\u5730\u5c06\u4efb\u610f\u8fde\u7eed\u548c\u5206\u7c7b\u7279\u5f81\u6dfb\u52a0\u5230\u6a21\u578b\u4e2d[9]\uff0c2017\u5e74IJCAI\u4f1a\u8bae\u4e2d\u63d0\u51fa\u7684\u5229\u7528LSTM\u8fdb\u884c\u5e8f\u5217\u63a8\u8350[6]\u3002<\/p>\n<p>2017\u5e74\u643a\u7a0b\u4e2a\u6027\u5316\u56e2\u961f\u5728AAAI\u4f1a\u8bae\u4e0a\u63d0\u51fa\u7684\u6df1\u5ea6\u6a21\u578baSDAE\uff0c\u901a\u8fc7\u5c06\u9644\u52a0\u7684 side information \u96c6\u6210\u5230\u8f93\u5165\u4e2d\uff0c\u53ef\u4ee5\u6539\u5584\u6570\u636e\u7a00\u758f\u548c\u51b7\u542f\u52a8\u95ee\u9898[4]\u3002<\/p>\n<p>\u5bf9\u4e8e\u53ec\u56de\u9636\u6bb5\u5f97\u5230\u7684\u5019\u9009\u96c6\uff0c\u4f1a\u5bf9\u5176\u8fdb\u884c\u66f4\u52a0\u590d\u6742\u548c\u7cbe\u786e\u7684\u6253\u5206\u4e0e\u91cd\u6392\u5e8f\uff0c\u8fdb\u800c\u5f97\u5230\u4e00\u4e2a\u66f4\u5c0f\u7684\u7528\u6237\u53ef\u80fd\u611f\u5174\u8da3\u7684\u4ea7\u54c1\u5217\u8868\u3002<\/p>\n<p>\u643a\u7a0b\u7684\u63a8\u8350\u6392\u5e8f\u5e76\u4e0d\u5355\u7eaf\u8ffd\u6c42\u70b9\u51fb\u7387\u6216\u8005\u8f6c\u5316\u7387\uff0c\u8fd8\u9700\u8981\u8003\u8651\u8ddd\u79bb\u63a7\u5236\uff0c\u4ea7\u54c1\u8d28\u91cf\u63a7\u5236\u7b49\u56e0\u7d20\u3002<\/p>\n<p>\u76f8\u6bd4\u9002\u7528\u4e8e\u641c\u7d22\u6392\u5e8f\uff0c\u6587\u672c\u76f8\u5173\u6027\u68c0\u7d22\u7b49\u9886\u57df\u7684 pairwise \u548c listwise \u65b9\u6cd5\uff0cpointwise \u65b9\u6cd5\u53ef\u4ee5\u901a\u8fc7\u53e0\u52a0\u5176\u4ed6\u63a7\u5236\u9879\u8fdb\u884c\u5e72\u9884\uff0c\u9002\u7528\u4e8e\u591a\u76ee\u6807\u4f18\u5316\u95ee\u9898\u3002<\/p>\n<p>\u5de5\u4e1a\u754c\u7684\u63a8\u8350\u65b9\u6cd5\u7ecf\u5386\u4ece\u7ebf\u6027\u6a21\u578b\uff0b\u5927\u91cf\u4eba\u5de5\u7279\u5f81\u5de5\u7a0b[11] -&gt; \u590d\u6742\u975e\u7ebf\u6027\u6a21\u578b-&gt; \u6df1\u5ea6\u5b66\u4e60\u7684\u53d1\u5c55\u3002Microsoft \u9996\u5148\u4e8e2007\u5e74\u63d0\u51fa\u91c7\u7528 Logistic Regression \u6765\u9884\u4f30\u641c\u7d22\u5e7f\u544a\u7684\u70b9\u51fb\u7387[12]\uff0c\u5e76\u4e8e\u540c\u5e74\u63d0\u51fa OWLQN \u4f18\u5316\u7b97\u6cd5\u7528\u4e8e\u6c42\u89e3\u5e26L1\u6b63\u5219\u7684LR\u95ee\u9898[13]\uff0c\u4e4b\u540e\u4e8e2010\u5e74\u63d0\u51fa\u57fa\u4e8eL2\u6b63\u5219\u7684\u5728\u7ebf\u5b66\u4e60\u7248\u672cAd Predictor[14]\u3002<\/p>\n<p>Google \u57282013\u5e74\u63d0\u51fa\u57fa\u4e8eL1\u6b63\u5219\u5316\u7684LR\u4f18\u5316\u7b97\u6cd5FTRL-Proximal[15]\u30022010\u5e74\u63d0\u51fa\u7684 Factorization Machine \u7b97\u6cd5[17] \u548c\u8fdb\u4e00\u6b652014\u5e74\u63d0\u51fa\u7684 Filed-aware Factorization Machine[18] \u65e8\u5728\u89e3\u51b3\u7a00\u758f\u6570\u636e\u4e0b\u7684\u7279\u5f81\u7ec4\u5408\u95ee\u9898\uff0c\u4ece\u800c\u907f\u514d\u91c7\u7528LR\u65f6\u9700\u8981\u7684\u5927\u91cf\u4eba\u5de5\u7279\u5f81\u7ec4\u5408\u5de5\u4f5c\u3002<\/p>\n<p>\u963f\u91cc\u4e8e2011\u5e74\u63d0\u51fa Mixture of Logistic Regression \u76f4\u63a5\u5728\u539f\u59cb\u7a7a\u95f4\u5b66\u4e60\u7279\u5f81\u4e4b\u95f4\u7684\u975e\u7ebf\u6027\u5173\u7cfb[19]\u3002<\/p>\n<p>Facebook \u4e8e2014\u5e74\u63d0\u51fa\u91c7\u7528 GBDT \u505a\u81ea\u52a8\u7279\u5f81\u7ec4\u5408\uff0c\u540c\u65f6\u878d\u5408Logistic Regression[20]\u3002<\/p>\n<p>\u8fd1\u5e74\u6765\uff0c\u6df1\u5ea6\u5b66\u4e60\u4e5f\u88ab\u6210\u529f\u5e94\u7528\u4e8e\u63a8\u8350\u6392\u5e8f\u9886\u57df\u3002Google \u57282016\u5e74\u63d0\u51fawide and deep learning \u65b9\u6cd5[21]\uff0c\u7efc\u5408\u6a21\u578b\u7684\u8bb0\u5fc6\u548c\u6cdb\u5316\u80fd\u529b\u3002<\/p>\n<p>\u8fdb\u4e00\u6b65\u534e\u4e3a\u63d0\u51fa DeepFM[15] \u6a21\u578b\u7528\u4e8e\u66ff\u6362wdl\u4e2d\u7684\u4eba\u5de5\u7279\u5f81\u7ec4\u5408\u90e8\u5206\u3002\u963f\u91cc\u57282017\u5e74\u5c06 attention \u673a\u5236\u5f15\u5165\uff0c\u63d0\u51fa Deep Interest Network[23]\u3002<\/p>\n<p>\u643a\u7a0b\u5728\u5b9e\u8df5\u76f8\u5e94\u7684\u6a21\u578b\u4e2d\u79ef\u7d2f\u4e86\u4e00\u5b9a\u7684\u7ecf\u9a8c\uff0c\u65e0\u8bba\u662f\u6700\u5e38\u7528\u7684\u903b\u8f91\u56de\u5f52\u6a21\u578b\uff08Logistic Regression\uff09\uff0c\u6811\u6a21\u578b\uff08GBDT\uff0cRandom Forest\uff09[16]\uff0c\u56e0\u5b50\u5206\u89e3\u673a\uff08FactorizationMachine\uff09\uff0c\u4ee5\u53ca\u8fd1\u671f\u63d0\u51fa\u7684 wdl \u6a21\u578b\u3002<\/p>\n<p>\u540c\u65f6\uff0c\u6211\u4eec\u8ba4\u4e3a\u5373\u4f7f\u5728\u6df1\u5ea6\u5b66\u4e60\u5927\u884c\u5176\u9053\u7684\u4eca\u4e0b\uff0c\u7cbe\u7ec6\u5316\u7684\u7279\u5f81\u5de5\u7a0b\u4ecd\u7136\u662f\u4e0d\u53ef\u6216\u7f3a\u7684\u3002<\/p>\n<p>\u57fa\u4e8e\u6392\u5e8f\u540e\u7684\u5217\u8868\uff0c\u5728\u7efc\u5408\u8003\u8651\u591a\u6837\u6027\u3001\u65b0\u9896\u6027\u3001Exploit &amp; Explore\u7b49\u56e0\u7d20\u540e\uff0c\u751f\u6210\u6700\u7ec8\u7684\u63a8\u8350\u7ed3\u679c\u3002<\/p>\n<p>\u672c\u6587\u4e4b\u540e\u5c06\u7740\u91cd\u4ecb\u7ecd\u53ec\u56de\u4e0e\u6392\u5e8f\u76f8\u5173\u7684\u5de5\u4f5c\u4e0e\u5b9e\u8df5\u3002<\/p>\n<p>\u4e00\u3001\u6570\u636e<\/p>\n<p>\u673a\u5668\u5b66\u4e60\uff1d\u6570\u636e\uff0b\u7279\u5f81\uff0b\u6a21\u578b<\/p>\n<p>\u5728\u4ecb\u7ecd\u53ec\u56de\u548c\u6392\u5e8f\u4e4b\u524d\uff0c\u5148\u7b80\u5355\u7684\u4e86\u89e3\u4e00\u4e0b\u6240\u7528\u5230\u7684\u6570\u636e\u3002\u643a\u7a0b\u4f5c\u4e3a\u5927\u578b OTA \u4f01\u4e1a\uff0c\u6bcf\u5929\u90fd\u6709\u6d77\u91cf\u7528\u6237\u6765\u8bbf\u95ee\uff0c\u79ef\u7d2f\u4e86\u5927\u91cf\u7684\u4ea7\u54c1\u6570\u636e\u4ee5\u53ca\u7528\u6237\u884c\u4e3a\u76f8\u5173\u7684\u6570\u636e\u3002\u5b9e\u9645\u5728\u53ec\u56de\u548c\u6392\u5e8f\u7684\u8fc7\u7a0b\u4e2d\u5927\u81f4\u4f7f\u7528\u5230\u4e86\u4ee5\u4e0b\u8fd9\u4e9b\u6570\u636e\uff1a<\/p>\n<p>\u4ea7\u54c1\u5c5e\u6027\uff1a\u4ea7\u54c1\u7684\u4e00\u4e9b\u56fa\u6709\u5c5e\u6027\uff0c\u5982\u9152\u5e97\u7684\u4f4d\u7f6e\uff0c\u661f\u7ea7\uff0c\u623f\u578b\u7b49\u3002<\/p>\n<p>\u4ea7\u54c1\u7edf\u8ba1\uff1a\u6bd4\u5982\u4ea7\u54c1\u4e00\u6bb5\u65f6\u95f4\u5185\u7684\u8ba2\u5355\u91cf\uff0c\u6d4f\u89c8\u91cf\uff0c\u641c\u7d22\u91cf\uff0c\u70b9\u51fb\u7387\u7b49\u3002<\/p>\n<p>\u7528\u6237\u753b\u50cf\uff1a\u7528\u6237\u57fa\u7840\u5c5e\u6027\uff0c\u6bd4\u5982\u5e74\u7eaa\uff0c\u6027\u522b\uff0c\u504f\u597d\u7b49\u7b49\u3002<\/p>\n<p>\u7528\u6237\u884c\u4e3a\uff1a\u7528\u6237\u7684\u8bc4\u8bba\uff0c\u8bc4\u5206\uff0c\u6d4f\u89c8\uff0c\u641c\u7d22\uff0c\u4e0b\u5355\u7b49\u884c\u4e3a\u3002<\/p>\n<p>\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0c\u9488\u5bf9\u7edf\u8ba1\u7c7b\u4fe1\u606f\uff0c\u53ef\u80fd\u9700\u8981\u8fdb\u884c\u4e00\u4e9b\u5e73\u6ed1\u3002\u4f8b\u5982\u9488\u5bf9\u5386\u53f2 CTR \u53cd\u9988\uff0c\u5229\u7528\u8d1d\u53f6\u65af\u5e73\u6ed1\u6765\u9884\u5904\u7406\u3002<\/p>\n<p>\u4e8c\u3001\u53ec\u56de<\/p>\n<p>\u53ec\u56de\u9636\u6bb5\u662f\u63a8\u8350\u6d41\u7a0b\u57fa\u7840\u7684\u4e00\u6b65\uff0c\u4ece\u6210\u5343\u4e0a\u4e07\u7684Item\u4e2d\u751f\u6210\u6570\u91cf\u6709\u9650\u7684\u5019\u9009\u96c6\uff0c\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u51b3\u5b9a\u4e86\u6392\u5e8f\u9636\u6bb5\u7684\u6548\u7387\u548c\u63a8\u8350\u7ed3\u679c\u7684\u4f18\u52a3\u3002<\/p>\n<p>\u800c\u7531OTA\u7684\u5c5e\u6027\u51b3\u5b9a\uff0c\u7528\u6237\u7684\u8bbf\u95ee\u884c\u4e3a\u5927\u591a\u662f\u4f4e\u9891\u7684\u3002\u8fd9\u5c31\u4f7f\u5f97user-item\u7684\u4ea4\u4e92\u6570\u636e\u662f\u6781\u5176\u7a00\u758f\u7684\uff0c\u8fd9\u5bf9\u53ec\u56de\u63d0\u51fa\u4e86\u5f88\u5927\u7684\u6311\u6218\u3002<\/p>\n<p>\u5728\u4e1a\u52a1\u5b9e\u8df5\u4e2d\uff0c\u6211\u4eec\u7ed3\u5408\u73b0\u6709\u7684\u901a\u7528\u63a8\u8350\u65b9\u6cd5\u548c\u4e1a\u52a1\u573a\u666f\uff0c\u7b5b\u9009\u548c\u6478\u7d22\u51fa\u4e86\u51e0\u79cd\u884c\u4e4b\u6709\u6548\u7684\u65b9\u6cd5\uff1a<\/p>\n<p>Real-timeIntention<\/p>\n<p>\u6211\u4eec\u7684\u5b9e\u65f6\u610f\u56fe\u7cfb\u7edf\u53ef\u4ee5\u6839\u636e\u7528\u6237\u6700\u8fd1\u6d4f\u89c8\u4e0b\u5355\u7b49\u884c\u4e3a\uff0c\u57fa\u4e8e\u9a6c\u5c14\u79d1\u592b\u9884\u6d4b\u6a21\u578b\u63a8\u8350\u6216\u8005\u4ea4\u53c9\u63a8\u8350\u51fa\u7684\u4ea7\u54c1\u3002\u8fd9\u4e9b\u5019\u9009\u4ea7\u54c1\u53ef\u4ee5\u6bd4\u8f83\u7cbe\u51c6\u7684\u53cd\u5e94\u51fa\u7528\u6237\u6700\u8fd1\u6700\u65b0\u7684\u610f\u613f\u3002<\/p>\n<p>BusinessRules<\/p>\n<p>\u4e1a\u52a1\u89c4\u5219\u662f\u8ba4\u4e3a\u8bbe\u5b9a\u7684\u89c4\u5219\uff0c\u7528\u6765\u9650\u5b9a\u63a8\u8350\u7684\u5185\u5bb9\u8303\u56f4\u7b49\u3002\u4f8b\u5982\u673a\u7968\u63a8\u9152\u5e97\u7684\u573a\u666f\uff0c\u9700\u8981\u901a\u8fc7\u4e1a\u52a1\u89c4\u5219\u6765\u9650\u5b9a\u63a8\u8350\u7684\u4ea7\u54c1\u53ea\u80fd\u662f\u9152\u5e97\uff0c\u800c\u4e0d\u4f1a\u63a8\u8350\u5176\u4ed6\u65c5\u6e38\u4ea7\u54c1\u3002<\/p>\n<p>Context-Based<\/p>\n<p>\u57fa\u4e8e Context \u7684\u63a8\u8350\u573a\u666f\u548c Context \u672c\u8eab\u5bc6\u5207\u76f8\u5173\uff0c\u4f8b\u5982\u4e0e\u5b63\u5019\u76f8\u5173\u7684\u65c5\u6e38\u4ea7\u54c1\uff08\u51ac\u5b63\u6ed1\u96ea\u3001\u5143\u65e6\u8de8\u5e74\u7b49\uff09\u3002<\/p>\n<p><img decoding=\"async\" class=\"alignnone size-full wp-image-9916\" src=\"https:\/\/www.growthhk.cn\/wp-content\/uploads\/mmexport1524568300468.png\" alt=\"\" width=\"538\" height=\"408\" \/><\/p>\n<p>LBS<\/p>\n<p>\u57fa\u4e8e\u7528\u6237\u7684\u5f53\u524d\u4f4d\u7f6e\u4fe1\u606f\uff0c\u7b5b\u9009\u51fa\u7684\u5468\u8fb9\u9152\u5e97\uff0c\u666f\u70b9\uff0c\u7f8e\u98df\u7b49\u7b49\uff0c\u6bd4\u8f83\u9002\u7528\u4e8e\u884c\u4e2d\u573a\u666f\u7684\u63a8\u8350\u3002<\/p>\n<p>\u5730\u7406\u4f4d\u7f6e\u8ddd\u79bb\u901a\u8fc7 GeoHash \u7b97\u6cd5\u8ba1\u7b97\uff0c\u5c06\u533a\u57df\u9012\u5f52\u5212\u5206\u4e3a\u89c4\u5219\u77e9\u5f62\uff0c\u5e76\u5bf9\u6bcf\u4e2a\u77e9\u5f62\u8fdb\u884c\u7f16\u7801\uff0c\u7b5b\u9009 GeoHash \u7f16\u7801\u76f8\u4f3c\u7684 POI\uff0c\u7136\u540e\u8fdb\u884c\u5b9e\u9645\u8ddd\u79bb\u8ba1\u7b97\u3002<\/p>\n<p>CollaborativeFiltering<\/p>\n<p>\u534f\u540c\u8fc7\u6ee4\u7b97\u6cd5\u662f\u63a8\u8350\u7cfb\u7edf\u5e7f\u6cdb\u4f7f\u7528\u7684\u4e00\u79cd\u89e3\u51b3\u5b9e\u9645\u95ee\u9898\u7684\u65b9\u6cd5\u3002\u643a\u7a0b\u4e2a\u6027\u5316\u56e2\u961f\u5728\u6df1\u5ea6\u5b66\u4e60\u4e0e\u63a8\u8350\u7cfb\u7edf\u7ed3\u5408\u7684\u9886\u57df\u8fdb\u884c\u4e86\u76f8\u5173\u7684\u7814\u7a76\u4e0e\u5e94\u7528\uff0c\u901a\u8fc7\u6539\u8fdb\u73b0\u6709\u7684\u6df1\u5ea6\u6a21\u578b\uff0c\u63d0\u51fa\u4e86\u4e00\u79cd\u6df1\u5ea6\u6a21\u578b aSDAE\u3002<\/p>\n<p>\u8be5\u6df7\u5408\u534f\u540c\u8fc7\u6ee4\u6a21\u578b\u662f SDAE \u7684\u4e00\u79cd\u53d8\u4f53\uff0c\u901a\u8fc7\u5c06\u9644\u52a0\u7684 side information \u96c6\u6210\u5230\u8f93\u5165\u4e2d\uff0c\u53ef\u4ee5\u6539\u5584\u6570\u636e\u7a00\u758f\u548c\u51b7\u542f\u52a8\u95ee\u9898\uff0c\u8be6\u60c5\u53ef\u4ee5\u53c2\u89c1\u6587\u732e[4]\u3002<\/p>\n<p>SequentialModel<\/p>\n<p>\u73b0\u6709\u7684\u77e9\u9635\u5206\u89e3(Matrix Factorization)\u65b9\u6cd5\u57fa\u4e8e\u5386\u53f2\u7684 user-item \u4ea4\u4e92\u5b66\u4e60\u7528\u6237\u7684\u957f\u671f\u5174\u8da3\u504f\u597d\uff0cMarkov chain \u901a\u8fc7\u5b66\u4e60 item \u95f4\u7684 transition graph \u5bf9\u7528\u6237\u7684\u5e8f\u5217\u884c\u4e3a\u5efa\u6a21[3]\u3002<\/p>\n<p>\u4e8b\u5b9e\u4e0a\uff0c\u5728\u65c5\u6e38\u573a\u666f\u4e0b\uff0c\u52a0\u5165\u7528\u6237\u884c\u4e3a\u7684\u5148\u540e\u987a\u5e8f\uff0c\u4ece\u800c\u80fd\u66f4\u597d\u7684\u53cd\u6620\u7528\u6237\u7684\u51b3\u7b56\u8fc7\u7a0b\u3002\u6211\u4eec\u7ed3\u5408 Matrix Factorization \u548c Markov chain \u4e3a\u6bcf\u4e2a\u7528\u6237\u6784\u5efa\u4e2a\u6027\u5316\u8f6c\u79fb\u77e9\u9635\uff0c\u4ece\u800c\u57fa\u4e8e\u7528\u6237\u7684\u5386\u53f2\u884c\u4e3a\u6765\u9884\u6d4b\u7528\u6237\u7684\u4e0b\u4e00\u884c\u4e3a\u3002<\/p>\n<p>\u5728\u65c5\u6e38\u573a\u666f\u4e2d\uff0c\u53ef\u4ee5\u7528\u6765\u9884\u6d4b\u7528\u6237\u4e0b\u4e00\u4e2a\u76ee\u7684\u5730\u6216\u8005 POI\u3002<\/p>\n<p>\u9664\u6b64\u4e4b\u5916\uff0c\u4e5f\u53ef\u4ee5\u4f7f\u7528 RNN \u6765\u8fdb\u884c\u5e8f\u5217\u63a8\u8350\uff0c\u6bd4\u5982\u57fa\u4e8e Session \u7684\u63a8\u8350[5]\uff0c\u4f7f\u7528\u8003\u8651\u65f6\u95f4\u95f4\u9694\u4fe1\u606f\u7684 LSTM \u6765\u505a\u4e0b\u4e00\u4e2a item \u7684\u63a8\u8350\u7b49[6]\u3002<\/p>\n<p>\u6b64\u5916\uff0c\u4e00\u4e9b\u5e38\u89c1\u7684\u6df1\u5ea6\u6a21\u578b(DNN, AE,CNN\u7b49)[7][8][9][10]\u90fd\u53ef\u4ee5\u5e94\u7528\u4e8e\u63a8\u8350\u7cfb\u7edf\u4e2d\uff0c\u4f46\u662f\u9488\u5bf9\u4e0d\u540c\u9886\u57df\u7684\u63a8\u8350\uff0c\u9700\u8981\u66f4\u591a\u7684\u9ad8\u6548\u7684\u6a21\u578b\u3002<\/p>\n<p>\u968f\u7740\u6df1\u5ea6\u5b66\u4e60\u6280\u672f\u7684\u53d1\u5c55\uff0c\u76f8\u4fe1\u6df1\u5ea6\u5b66\u4e60\u5c06\u4f1a\u6210\u4e3a\u63a8\u8350\u7cfb\u7edf\u9886\u57df\u4e2d\u4e00\u9879\u975e\u5e38\u91cd\u8981\u7684\u6280\u672f\u624b\u6bb5\u3002<\/p>\n<p>\u4ee5\u4e0a\u51e0\u79cd\u7c7b\u578b\u7684\u53ec\u56de\u65b9\u6cd5\u5404\u6709\u4f18\u52bf\uff0c\u5728\u5b9e\u8df5\u4e2d\uff0c\u9488\u5bf9\u4e0d\u540c\u573a\u666f\uff0c\u7ed3\u5408\u4f7f\u7528\u591a\u79cd\u65b9\u6cd5\uff0c\u63d0\u4f9b\u7ed9\u7528\u6237\u6700\u4f73\u7684\u63a8\u8350\uff0c\u4ee5\u6b64\u63d0\u5347\u7528\u6237\u4f53\u9a8c\uff0c\u589e\u52a0\u7528\u6237\u7c98\u6027\u3002<\/p>\n<p>\u4e09\u3001\u6392\u5e8f<\/p>\n<p>\u4ee5\u5de5\u4e1a\u754c\u5728\u5e7f\u544a\u3001\u641c\u7d22\u3001\u63a8\u8350\u7b49\u9886\u57df\u7684\u5b9e\u8df5\u7ecf\u9a8c\uff0c\u5728\u6570\u636e\u7ed9\u5b9a\u7684\u6761\u4ef6\u4e0b\uff0c\u7ecf\u5386\u4e86\u4ece\u7b80\u5355\u7ebf\u6027\u6a21\u578b\uff0b\u5927\u91cf\u4eba\u5de5\u7279\u5f81\u5de5\u7a0b\u5230\u590d\u6742\u975e\u7ebf\u6027\u6a21\u578b\uff0b\u81ea\u52a8\u7279\u5f81\u5b66\u4e60\u7684\u6f14\u53d8\u3002<\/p>\n<p>\u5728\u6784\u5efa\u643a\u7a0b\u4e2a\u6027\u5316\u63a8\u8350\u7cfb\u7edf\u7684\u5b9e\u8df5\u8fc7\u7a0b\u4e2d\uff0c\u5bf9\u4e8e\u63a8\u8350\u6392\u5e8f\u8fd9\u4e2a\u7279\u5b9a\u95ee\u9898\u6709\u4e00\u4e9b\u81ea\u5df1\u7684\u601d\u8003\u548c\u603b\u7ed3\uff0c\u5e76\u5c06\u4ece\u7279\u5f81\u548c\u6a21\u578b\u8fd9\u4e24\u65b9\u9762\u5c55\u5f00\u3002<\/p>\n<p>Model<\/p>\n<p>\u4e2a\u6027\u5316\u6392\u5e8f\u6a21\u578b\u65e8\u5728\u5229\u7528\u6bcf\u4e2a\u7528\u6237\u7684\u5386\u53f2\u884c\u4e3a\u6570\u636e\u96c6\u5efa\u7acb\u5176\u5404\u81ea\u7684\u6392\u5e8f\u6a21\u578b\uff0c\u672c\u8d28\u4e0a\u53ef\u4ee5\u770b\u4f5c\u591a\u4efb\u52a1\u5b66\u4e60(multi-task learning)\u3002<\/p>\n<p>\u4e8b\u5b9e\u4e0a\uff0c\u901a\u8fc7\u52a0\u5165 conjunctionfeatures\uff0c\u4e5f\u5c31\u662f\u52a0\u5165 user \u548c product \u7684\u4ea4\u53c9\u7279\u5f81\uff0c\u53ef\u4ee5\u5c06\u7279\u5b9a\u7684 multi-task \u4efb\u52a1\u7b80\u5316\u4e3a\u5355\u4efb\u52a1\u6a21\u578b\u3002<\/p>\n<p>\u68b3\u7406\u5de5\u4e1a\u754c\u5e94\u7528\u7684\u6392\u5e8f\u6a21\u578b\uff0c\u5927\u81f4\u7ecf\u5386\u4e09\u4e2a\u9636\u6bb5\uff0c\u5982\u4e0b\u56fe\u6240\u793a\uff1a<\/p>\n<p><img decoding=\"async\" class=\"alignnone size-full wp-image-9917\" src=\"https:\/\/www.growthhk.cn\/wp-content\/uploads\/mmexport1524568358371.jpg\" alt=\"\" width=\"951\" height=\"529\" \/><\/p>\n<p>\u672c\u6587\u5e76\u4e0d\u51c6\u5907\u8be6\u7ec6\u4ecb\u7ecd\u4e0a\u56fe\u4e2d\u7684\u7b97\u6cd5\u7ec6\u8282\uff0c\u611f\u5174\u8da3\u7684\u8bfb\u8005\u53ef\u4ee5\u67e5\u770b\u76f8\u5173\u8bba\u6587\uff0c\u4ee5\u4e0b\u51e0\u70b9\u662f\u6211\u4eec\u7684\u4e00\u4e9b\u5b9e\u8df5\u7ecf\u9a8c\u548c\u4f53\u4f1a\u3002<\/p>\n<ul>\n<li>\u5728\u5b9e\u8df5\u4e2d\u9009\u7528\u4ee5LR\u4e3a\u4e3b\u7684\u6a21\u578b\uff0c\u901a\u8fc7\u5bf9\u6570\u636e\u79bb\u6563\u5316\u3001\u5206\u5e03\u8f6c\u6362\u7b49\u975e\u7ebf\u6027\u5904\u7406\u540e\u4f7f\u7528LR\u3002\u4e00\u822c\u7684\uff0c\u91c7\u7528L1\u6b63\u5219\u4fdd\u8bc1\u6a21\u578b\u6743\u91cd\u7684\u7a00\u758f\u6027\u3002<\/li>\n<li>\u5728\u4f18\u5316\u7b97\u6cd5\u7684\u9009\u62e9\u4e0a\uff0c\u4f7f\u7528 OWL-QN \u505a batch learning\uff0cFTRL \u505a online learning\u3002<\/li>\n<li>\u5b9e\u8df5\u4e2d\u5229\u7528\u56e0\u5b50\u5206\u89e3\u673a\uff08FactorizationMachine\uff09\u5f97\u5230\u7684\u7279\u5f81\u4ea4\u53c9\u7cfb\u6570\u6765\u9009\u62e9\u5582\u5165LR\u6a21\u578b\u7684\u4ea4\u53c9\u7279\u5f81\u7ec4\u5408\uff0c\u4ece\u800c\u907f\u514d\u4e86\u7e41\u6742\u7684\u7279\u5f81\u9009\u62e9\u5de5\u4f5c\u3002\u4e00\u822c\u7684\u53d7\u9650\u4e8e\u6a21\u578b\u590d\u6742\u5ea6\u53ea\u8fdb\u884c\u4e8c\u9636\u5c55\u5f00\u3002<\/li>\n<li>\u5bf9\u4e8e\u4e09\u9636\u4ee5\u4e0a\u7684\u7279\u5f81\u7ec4\u5408\u53ef\u4ee5\u5229\u7528\u57fa\u4e8e mutual information \u7b49\u65b9\u6cd5\u5904\u7406\u3002\u5df2\u6709\u9488\u5bf9\u9ad8\u9636\u56e0\u5b50\u5206\u89e3\u673a\uff08HighOrder FM\uff09\u7684\u7814\u7a76\uff0c\u53c2\u89c1\u6587\u732e[24]\u3002<\/li>\n<li>\u5bf9\u4e8e Wide and Deep Learning\uff0c\u5c06 wide \u90e8\u5206\u66ff\u6362 gbdt \u7ec4\u5408\u7279\u5f81\uff0c\u5728\u5b9e\u9a8c\u4e2d\u53d6\u5f97\u4e86\u8f83\u597d\u7684\u6548\u679c\uff0c\u5e76\u5c06\u5728\u8fd1\u671f\u4e0a\u7ebf\u3002<\/li>\n<li>\u540e\u7eed\u7684\u5de5\u4f5c\u5c06\u9488\u5bf9\u5982\u4f55\u8fdb\u884c wide \u90e8\u5206\u548c deep \u90e8\u5206\u7684 alternatingtraining \u5c55\u5f00\u3002<\/li>\n<li>Feature Engineering<\/li>\n<\/ul>\n<p>\u4e8b\u5b9e\u4e0a\uff0c\u867d\u7136\u6df1\u5ea6\u5b66\u4e60\u7b49\u65b9\u6cd5\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u51cf\u5c11\u4e86\u7e41\u6742\u7684\u7279\u5f81\u5de5\u7a0b\u5de5\u4f5c\uff0c\u4f46\u6211\u4eec\u8ba4\u4e3a\u7cbe\u5fc3\u8bbe\u8ba1\u7684\u7279\u5f81\u5de5\u7a0b\u4ecd\u65e7\u662f\u4e0d\u53ef\u6216\u7f3a\u7684, \u5176\u4e2d\u5982\u4f55\u8fdb\u884c\u7279\u5f81\u7ec4\u5408\u662f\u6211\u4eec\u5728\u5b9e\u8df5\u4e2d\u7740\u91cd\u8003\u8651\u7684\u95ee\u9898\u3002\u4e00\u822c\u7684\uff0c\u53ef\u4ee5\u5206\u4e3a\u663e\u5f0f\u7279\u5f81\u7ec4\u5408\u548c\u534a\u663e\u5f0f\u7279\u5f81\u7ec4\u5408\u3002<\/p>\n<p>\u663e\u5f0f\u7279\u5f81\u7ec4\u5408<\/p>\n<p>\u5bf9\u7279\u5f81\u8fdb\u884c\u79bb\u6563\u5316\u540e\u7136\u540e\u8fdb\u884c\u53c9\u4e58\uff0c\u91c7\u7528\u7b1b\u5361\u5c14\u79ef(cartesian product)\u3001\u5185\u79ef(inner product)\u7b49\u65b9\u5f0f\u3002<\/p>\n<p>\u5728\u6784\u9020\u4ea4\u53c9\u7279\u5f81\u7684\u8fc7\u7a0b\u4e2d\uff0c\u9700\u8981\u8fdb\u884c\u7279\u5f81\u79bb\u6563\u5316\uff1b\u9488\u5bf9\u4e0d\u540c\u7684\u7279\u5f81\u7c7b\u578b\uff0c\u6709\u4e0d\u540c\u7684\u5904\u7406\u65b9\u5f0f\u3002<\/p>\n<p>1. numerical feature<\/p>\n<p>\u65e0\u76d1\u7763\u79bb\u6563\u5316\uff1a\u6839\u636e\u7b80\u5355\u7edf\u8ba1\u91cf\u8fdb\u884c\u7b49\u9891\u3001\u7b49\u5bbd\u3001\u5206\u4f4d\u70b9\u7b49\u5212\u5206\u533a\u95f4<\/p>\n<p>\u6709\u76d1\u7763\u79bb\u6563\u5316\uff1a1R\u65b9\u6cd5\uff0cEntropy-BasedDiscretization (e.g. D2\uff0cMDLP)<\/p>\n<p>2. ordinal feature\uff08\u6709\u5e8f\u7279\u5f81\uff09<\/p>\n<p>\u7f16\u7801\u8868\u793a\u503c\u4e4b\u95f4\u7684\u987a\u5e8f\u5173\u7cfb\u3002\u6bd4\u5982\u5bf9\u4e8e\u536b\u751f\u6761\u4ef6\u8fd9\u4e00\u7279\u5f81\uff0c\u5206\u522b\u6709\u5dee\uff0c\u4e2d\uff0c\u597d\u4e09\u6863\uff0c\u90a3\u4e48\u53ef\u4ee5\u5206\u522b\u7f16\u7801\u4e3a(1,0,0),(1,1,0),(1,1,1)\u3002<\/p>\n<p>3. categorical feature (\u65e0\u5e8f\u7279\u5f81)<\/p>\n<p><img decoding=\"async\" class=\"alignnone size-full wp-image-9918\" src=\"https:\/\/www.growthhk.cn\/wp-content\/uploads\/Screenshot_2018-04-24-19-14-17-053_\u5fae\u4fe1.png\" alt=\"\" width=\"1024\" height=\"512\" \/><\/p>\n<p>\u79bb\u6563\u5316\u4e3a\u54d1\u53d8\u91cf\uff0c\u5c06\u4e00\u7ef4\u4fe1\u606f\u5d4c\u5165\u6a21\u578b\u7684bias\u4e2d\uff0c\u8d77\u5230\u7b80\u5316\u903b\u8f91\u56de\u5f52\u6a21\u578b\u7684\u4f5c\u7528\uff0c\u964d\u4f4e\u4e86\u6a21\u578b\u8fc7\u62df\u5408\u7684\u98ce\u9669\u3002<\/p>\n<p>\u79bb\u6563\u7279\u5f81\u7ecf\u8fc7OHE\u540e\uff0c\u6bcf\u4e2a\u5206\u7c7b\u578b\u53d8\u91cf\u7684\u5404\u4e2a\u503c\u5728\u6a21\u578b\u4e2d\u90fd\u53ef\u4ee5\u770b\u4f5c\u72ec\u7acb\u53d8\u91cf\uff0c\u589e\u5f3a\u62df\u5408\u80fd\u529b\u3002\u4e00\u822c\u7684\uff0c\u5f53\u6a21\u578b\u52a0\u6b63\u5219\u5316\u7684\u60c5\u51b5\u4e0b\u7ea6\u675f\u6a21\u578b\u81ea\u7531\u5ea6\uff0c\u6211\u4eec\u8ba4\u4e3aOHE\u66f4\u597d\u3002<\/p>\n<p>\u5229\u7528 feature hash \u6280\u672f\u5c06\u9ad8\u7ef4\u7a00\u758f\u7279\u5f81\u6620\u5c04\u5230\u56fa\u5b9a\u7ef4\u5ea6\u7a7a\u95f4<\/p>\n<p>\u534a\u663e\u5f0f\u7279\u5f81\u7ec4\u5408<\/p>\n<p>\u533a\u522b\u4e8e\u663e\u5f0f\u7279\u5f81\u7ec4\u5408\u5177\u6709\u660e\u786e\u7684\u7ec4\u5408\u89e3\u91ca\u4fe1\u606f\uff0c\u534a\u663e\u5f0f\u7279\u5f81\u7ec4\u5408\u901a\u5e38\u7684\u505a\u6cd5\u662f\u57fa\u4e8e\u6811\u65b9\u6cd5\u5f62\u6210\u7279\u5f81\u5212\u5206\u5e76\u7ed9\u51fa\u76f8\u5e94\u7ec4\u5408\u8def\u5f84\u3002<\/p>\n<p>\u4e00\u822c\u505a\u6cd5\u662f\u5c06\u6837\u672c\u7684\u8fde\u7eed\u503c\u7279\u5f81\u8f93\u5165 ensemble tree\uff0c\u5206\u522b\u5728\u6bcf\u9897\u51b3\u7b56\u6811\u6cbf\u7740\u7279\u5b9a\u5206\u652f\u8def\u5f84\u6700\u7ec8\u843d\u5165\u67d0\u4e2a\u53f6\u5b50\u7ed3\u70b9\u5f97\u5230\u5176\u7f16\u53f7\uff0c\u672c\u8d28\u4e0a\u662f\u8fd9\u4e9b\u7279\u5f81\u5728\u7279\u5b9a\u53d6\u503c\u533a\u95f4\u5185\u7684\u7ec4\u5408\u3002<\/p>\n<p>ensemble tree \u53ef\u4ee5\u91c7\u7528 Gbdt \u6216\u8005 random forest \u5b9e\u73b0\u3002\u6bcf\u4e00\u8f6e\u8fed\u4ee3\uff0c\u4ea7\u751f\u4e00\u68f5\u65b0\u6811\uff0c\u6700\u7ec8\u901a\u8fc7 one-hotencoding \u8f6c\u5316\u4e3a binary vector\uff0c\u5982\u4e0b\u56fe\u6240\u793a\u3002<\/p>\n<p><img decoding=\"async\" class=\"alignnone size-full wp-image-9919\" src=\"https:\/\/www.growthhk.cn\/wp-content\/uploads\/mmexport1524568563955.jpg\" alt=\"\" width=\"584\" height=\"334\" \/><\/p>\n<p>\u4ee5\u4e0b\u51e0\u70b9\u662f\u6211\u4eec\u5728\u5b9e\u8df5\u4e2d\u7684\u4e00\u4e9b\u603b\u7ed3\u548c\u601d\u8003\u3002<\/p>\n<p>\u5728\u5b9e\u9a8c\u4e2d\u53d1\u73b0\u5982\u679c\u5c06\u8fde\u7eed\u503c\u7279\u5f81\u8fdb\u884c\u79bb\u6563\u5316\u540e\u5582\u5165 gbdt\uff0cgbdt \u7684\u6548\u679c\u4e0d\u4f73\uff0cAUC \u6bd4\u8f83\u4f4e\u3002<\/p>\n<p>\u8fd9\u662f\u56e0\u4e3a gbdt \u672c\u8eab\u80fd\u5f88\u597d\u7684\u5904\u7406\u975e\u7ebf\u6027\u7279\u5f81\uff0c\u4f7f\u7528\u79bb\u6563\u5316\u540e\u7684\u7279\u5f81\u53cd\u800c\u6ca1\u4ec0\u4e48\u6548\u679c\u3002<\/p>\n<p>xgboost \u7b49\u6811\u6a21\u578b\u65e0\u6cd5\u6709\u6548\u5904\u7406\u9ad8\u7ef4\u7a00\u758f\u7279\u5f81\u6bd4\u5982 user id \u7c7b\u7279\u5f81\uff0c\u53ef\u4ee5\u91c7\u7528\u7684\u66ff\u4ee3\u65b9\u5f0f\u662f\uff1a\u5c06\u8fd9\u7c7b id \u5229\u7528\u4e00\u79cd\u65b9\u5f0f\u8f6c\u6362\u4e3a\u4e00\u4e2a\u6216\u591a\u4e2a\u65b0\u7684\u8fde\u7eed\u578b\u7279\u5f81\uff0c\u7136\u540e\u7528\u4e8e\u6a21\u578b\u8bad\u7ec3\u3002<\/p>\n<p>\u9700\u8981\u6ce8\u610f\u7684\u662f\u5f53\u91c7\u7528\u53f6\u5b50\u7ed3\u70b9\u7684 index \u4f5c\u4e3a\u7279\u5f81\u8f93\u51fa\u9700\u8981\u8003\u8651\u6bcf\u68f5\u6811\u7684\u53f6\u5b50\u7ed3\u70b9\u5e76\u4e0d\u5b8c\u5168\u540c\u5904\u4e8e\u76f8\u540c\u6df1\u5ea6\u3002<\/p>\n<p>\u5b9e\u8df5\u4e2d\u91c7\u7528\u4e86 Monte Carlo Search \u5bf9 xgboost \u7684\u4f17\u591a\u53c2\u6570\u8fdb\u884c\u8d85\u53c2\u6570\u9009\u62e9\u3002<\/p>\n<p>\u5728\u79bb\u7ebf\u8bad\u7ec3\u9636\u6bb5\u91c7\u7528\u57fa\u4e8e Spark \u96c6\u7fa4\u7684 xgboost \u5206\u5e03\u5f0f\u8bad\u7ec3\uff0c\u800c\u5728\u7ebf\u9884\u6d4b\u65f6\u5219\u5bf9\u6a21\u578b\u6587\u4ef6\u76f4\u63a5\u8fdb\u884c\u89e3\u6790\uff0c\u80fd\u591f\u6ee1\u8db3\u7ebf\u4e0a\u5b9e\u65f6\u54cd\u5e94\u7684\u9700\u6c42\u3002<\/p>\n<p>\u6b64\u5916\uff0c\u5728\u5b9e\u8df5\u53d1\u73b0\u5355\u7eaf\u91c7\u7528 xgboost \u81ea\u52a8\u5b66\u5230\u7684\u9ad8\u9636\u7ec4\u5408\u7279\u5f81\u540e\u7eed\u8f93\u5165LR\u6a21\u578b\u5e76\u4e0d\u80fd\u5b8c\u5168\u66ff\u4ee3\u4eba\u5de5\u7279\u5f81\u5de5\u7a0b\u7684\u4f5c\u7528\uff1b<\/p>\n<p>\u53ef\u4ee5\u5c06\u539f\u59cb\u7279\u5f81\u4ee5\u53ca\u4e00\u4e9b\u4eba\u5de5\u7ec4\u5408\u7684\u9ad8\u9636\u4ea4\u53c9\u7279\u5f81\u540c xgboost \u5b66\u4e60\u5230\u7684\u7279\u5f81\u7ec4\u5408\u4e00\u8d77\u653e\u5165\u540e\u7eed\u7684\u6a21\u578b\uff0c\u83b7\u5f97\u66f4\u597d\u7684\u6548\u679c\u3002<\/p>\n<p>\u56db\u3001\u603b\u7ed3<\/p>\n<p>\u5b8c\u6574\u7684\u63a8\u8350\u7cfb\u7edf\u662f\u4e00\u4e2a\u5e9e\u5927\u7684\u7cfb\u7edf\uff0c\u6d89\u53ca\u591a\u4e2a\u65b9\u9762\uff0c\u9664\u4e86\u53ec\u56de\u3001\u6392\u5e8f\u3001\u5217\u8868\u751f\u4ea7\u7b49\u6b65\u9aa4\u5916\uff0c\u8fd8\u6709\u6570\u636e\u51c6\u5907\u4e0e\u5904\u7406\uff0c\u5de5\u7a0b\u67b6\u6784\u4e0e\u5b9e\u73b0\uff0c\u524d\u7aef\u5c55\u73b0\u7b49\u7b49\u3002<\/p>\n<p>\u5728\u5b9e\u9645\u4e2d\uff0c\u901a\u8fc7\u628a\u8fd9\u4e9b\u6a21\u5757\u96c6\u6210\u5728\u4e00\u8d77\uff0c\u6784\u6210\u4e86\u4e00\u4e2a\u96c6\u56e2\u901a\u7528\u63a8\u8350\u7cfb\u7edf\uff0c\u5bf9\u5916\u63d0\u4f9b\u63a8\u670d\u52a1\uff0c\u5e94\u7528\u572810\u591a\u4e2a\u680f\u4f4d\uff0c60\u591a\u4e2a\u573a\u666f\uff0c\u53d6\u5f97\u4e86\u5f88\u597d\u7684\u6548\u679c\u3002<\/p>\n<p>\u672c\u6587\u4fa7\u91cd\u4ecb\u7ecd\u4e86\u53ec\u56de\u4e0e\u6392\u5e8f\u7b97\u6cd5\u76f8\u5173\u7684\u76ee\u524d\u5df2\u6709\u7684\u4e00\u4e9b\u5de5\u4f5c\u4e0e\u5b9e\u8df5\uff0c\u4e0b\u4e00\u6b65\uff0c\u8ba1\u5212\u5f15\u5165\u66f4\u591a\u5730\u6df1\u5ea6\u6a21\u578b\u6765\u5904\u7406\u53ec\u56de\u4e0e\u6392\u5e8f\u95ee\u9898\uff0c\u5e76\u7ed3\u5408\u5728\u7ebf\u5b66\u4e60\u3001\u5f3a\u5316\u5b66\u4e60\u3001\u8fc1\u79fb\u5b66\u4e60\u7b49\u65b9\u9762\u7684\u8fdb\u5c55\uff0c\u4f18\u5316\u63a8\u8350\u7684\u6574\u4f53\u8d28\u91cf\u3002<\/p>\n<p>References<\/p>\n<p>[1] Koren, Yehuda,Robert Bell, and Chris Volinsky. &#8220;Matrix <span class=\"wpcom_keyword_link\"><a href=\"https:\/\/www.growthhk.cn\/tag\/fact\" target=\"_blank\" title=\"fact\">fact<\/a><\/span>orization techniques forrecommender systems.&#8221; Computer 42.8 (2009).<br \/>[2] Sedhain, Suvash,et al. &#8220;Autorec: Autoencoders meet collaborative filtering.&#8221; Proceedingsof the 24th International Conference on World Wide Web. ACM, 2015.<br \/>[3] Rendle, Steffen,Christoph Freudenthaler, and Lars Schmidt-Thieme. &#8220;Factorizingpersonalized markov chains for next-basket recommendation.&#8221; Proceedings ofthe 19th international conference on World wide web. ACM, 2010.<br \/>[4] Dong, Xin, etal. &#8220;A Hybrid Collaborative Filtering Model with Deep Structure forRecommender Systems.&#8221; AAAI. 2017.<br \/>[5] Hidasi, Bal\u00e1zs,et al. &#8220;Session-based recommendations with recurrent neuralnetworks.&#8221; arXiv preprint arXiv:1511.06939 (2015).<br \/>[6] Zhu, Yu, et al.&#8221;What to Do Next: Modeling User Behaviors by Time-LSTM.&#8221; Proceedingsof the Twenty-Sixth International Joint Conference on Artificial Intelligence,IJCAI-17. 2017.<br \/>[7] Barkan, Oren,and Noam Koenigstein. &#8220;Item2vec: neural item embedding for collaborativefiltering.&#8221; Machine Learning for Signal Processing (MLSP), 2016 IEEE 26thInternational Workshop on. IEEE, 2016.<br \/>[8] Wang, Hao,Naiyan Wang, and Dit-Yan Yeung. &#8220;Collaborative deep learning forrecommender systems.&#8221; Proceedings of the 21th ACM SIGKDD InternationalConference on Knowledge Discovery and Data Mining. ACM, 2015.<br \/>[9] Covington, Paul,Jay Adams, and Emre Sargin. &#8220;Deep neural networks for youtuberecommendations.&#8221; Proceedings of the 10th ACM Conference on RecommenderSystems. ACM, 2016.<br \/>[10] Kim, Donghyun, et al. &#8220;Convolutional matrix factorization fordocument context-aware recommendation.&#8221; Proceedings of the 10th ACMConference on Recommender Systems. ACM, 2016.[11] https:\/\/mli.github.io\/2013\/03\/24\/the-end-of-feature-engineering-and-linear-model\/[12] Richardson, Matthew, Ewa Dominowska, and Robert Ragno.&#8221;Predicting clicks: estimating the click-through rate for new ads.&#8221;Proceedings of the 16th international conference on World Wide Web. ACM, 2007[13] Andrew, Galen, and Jianfeng Gao. &#8220;Scalable training of L1-regularized log-linear models.&#8221; Proceedings of the 24th internationalconference on Machine learning. ACM, 2007.[14] Graepel, Thore, et al. &#8220;Web-scale bayesian click-through rateprediction for sponsored search advertising in microsoft&#8217;s bing searchengine.&#8221; Omnipress, 2010.<br \/>[15] McMahan, H. Brendan, et al. &#8220;Ad click prediction: a view fromthe trenches.&#8221; Proceedings of the 19th ACM SIGKDD international conferenceon Knowledge discovery and data mining. ACM, 2013.<br \/>[16] Chen, Tianqi, and Carlos Guestrin. &#8220;Xgboost: A scalable treeboosting system.&#8221; Proceedings of the 22nd acm sigkdd internationalconference on knowledge discovery and data mining. ACM, 2016.<br \/>[17] Rendle, Steffen. &#8220;Factorization machines.&#8221; Data Mining(ICDM), 2010 IEEE 10th International Conference on. IEEE, 2010.<br \/>[18] Juan, Yuchin, et al. &#8220;Field-aware factorization machines forCTR prediction.&#8221; Proceedings of the 10th ACM Conference on RecommenderSystems. ACM, 2016.<br \/>[19] Gai, Kun, et al. &#8220;Learning Piece-wise Linear Models fromLarge Scale Data for Ad Click Prediction.&#8221; arXiv preprint arXiv:1704.05194(2017).<br \/>[20] He, Xinran, et al. &#8220;Practical lessons from predicting clickson ads at facebook.&#8221; Proceedings of the Eighth International Workshop onData Mining for Online Advertising. ACM, 2014.<br \/>[21] Cheng, Heng-Tze, et al. &#8220;Wide &amp; deep learning forrecommender systems.&#8221; Proceedings of the 1st Workshop on Deep Learning forRecommender Systems. ACM, 2016.<br \/>[22] Guo, Huifeng, et al. &#8220;DeepFM: A Factorization-Machine basedNeural Network for CTR Prediction.&#8221; arXiv preprint arXiv:1703.04247(2017).<br \/>[23] Zhou, Guorui, et al. &#8220;Deep Interest Network for Click-ThroughRate Prediction.&#8221; arXiv preprint arXiv:1706.06978 (2017).<br \/>[24] Blondel, Mathieu, et al. &#8220;Higher-orderfactorization machines.&#8221; Advances in Neural Information ProcessingSystems. 2016.<br \/>[25] http:\/\/breezedeus.github.io\/2014\/11\/20\/breezedeus-feature-hashing.html<br \/>[26] https:\/\/en.wikipedia.org\/wiki\/Categorical_variable<br \/>[27] https:\/\/www.zhihu.com\/question\/48674426<br \/>[28] \u591a\u9ad8\u7684AUC\u624d\u7b97\u9ad8\uff1fhttps:\/\/zhuanlan.zhihu.com\/p\/24217322<\/p>\n<p>\u6587\uff1a<span class=\"wpcom_tag_link\"><a href=\"https:\/\/www.growthhk.cn\/tag\/%e6%90%ba%e7%a8%8b%e6%8a%80%e6%9c%af%e4%b8%ad%e5%bf%83\" title=\"\u643a\u7a0b\u6280\u672f\u4e2d\u5fc3\" target=\"_blank\">\u643a\u7a0b\u6280\u672f\u4e2d\u5fc3<\/a><\/span><\/p>\n<p><span id=\"more-9912\"><\/span><strong>\u76f8\u5173\u6587\u7ae0\u63a8\u8350\uff1a<\/strong><\/p>\n<blockquote><p>\u300a<a href=\"https:\/\/www.growthhk.cn\/cgo\/9882.html\">\u300a\u589e\u957f\u9ed1\u5ba2\u300b| \u5bfb\u627e\u751f\u547d\u4e2d\u7684AHA\u65f6\u523b<\/a>\u300b<br \/>\u300a<a href=\"https:\/\/www.growthhk.cn\/cgo\/9861.html\">\u9999\u6c34\u4ea7\u54c1\u4e3a\u4f8b\uff0c\u9996\u5e2d\u589e\u957f\u5b98\u5e38\u7528\u6570\u636e\u91c7\u96c6\u3001\u5206\u6790\u3001\u6316\u6398\u7684\u8fc7\u7a0b\u53ca\u7b97\u6cd5\u6a21\u578b<\/a>\u300b<br \/>\u300a<a href=\"https:\/\/www.growthhk.cn\/cgo\/9912.html\">\u9996\u5e2d\u589e\u957f\u5b98\u7684\u5317\u6781\u661f\u6307\u6807\uff1a\u65b0\u7528\u6237\u6fc0\u6d3b<\/a>\u300b<\/p><\/blockquote>\n<p>\u66f4\u591a\u7cbe\u5f69\uff0c\u5173\u6ce8\uff1a<span class=\"wpcom_keyword_link\"><a href=\"https:\/\/www.growthhk.cn\/\" target=\"_blank\" title=\"\u589e\u957f\u9ed1\u5ba2\">\u589e\u957f\u9ed1\u5ba2<\/a><\/span>\uff08GrowthHK.cn\uff09<\/p>\n<p>\u589e\u957f\u9ed1\u5ba2\uff08Growth Hacker\uff09\u662f\u4f9d\u9760\u6280\u672f\u548c\u6570\u636e\u6765\u8fbe\u6210\u5404\u79cd\u8425\u9500\u76ee\u6807\u7684\u65b0\u578b\u56e2\u961f\u89d2\u8272\u3002\u4ece\u5355\u7ebf\u601d\u7ef4\u8005\u65f6\u5e38\u5ffd\u7565\u7684\u89d2\u5ea6\u548c\u9ad8\u5ea6\uff0c\u68b3\u7406\u6574\u5408\u4ea7\u54c1\u53d1\u5c55\u7684\u56e0\u7d20\uff0c\u5b9e\u73b0\u4f4e\u6210\u672c\u751a\u81f3\u96f6\u6210\u672c\u5e26\u6765\u7684\u6709\u6548\u589e\u957f\u2026<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u643a\u7a0b\u57fa\u7840\u4e1a\u52a1\u7814\u53d1\u90e8-\u6570\u636e\u4ea7\u54c1\u548c\u670d\u52a1\u7ec4\uff0c\u4e13\u6ce8\u4e8e\u4e2a\u6027\u5316\u63a8\u8350\u3001\u81ea\u7136\u8bed\u8a00\u5904\u7406\u3001\u56fe\u50cf\u8bc6\u522b\u7b49\u4eba\u5de5\u667a\u80fd\u9886\u57df\u7684\u5148\u8fdb\u6280\u672f\u5728\u65c5\u6e38\u884c\u4e1a\u7684\u5e94\u7528\u7814\u7a76\u5e76\u843d\u5730\u4ea7\u751f\u4ef7\u503c\u3002 \u76ee\u524d\uff0c\u56e2\u961f\u5df2\u7ecf\u4e3a\u643a\u7a0b\u63d0\u4f9b\u4e86\u901a\u7528\u5316\u7684\u4e2a\u6027\u5316\u2026<\/p>\n","protected":false},"author":1,"featured_media":9919,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[10],"tags":[2495,2494,70],"special":[],"class_list":["post-9914","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-cgo","tag-2495","tag-2494","tag-70","entry"],"views":144435,"_links":{"self":[{"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/posts\/9914","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/comments?post=9914"}],"version-history":[{"count":0,"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/posts\/9914\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/media\/9919"}],"wp:attachment":[{"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/media?parent=9914"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/categories?post=9914"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/tags?post=9914"},{"taxonomy":"special","embeddable":true,"href":"https:\/\/www.growthhk.cn\/wp-json\/wp\/v2\/special?post=9914"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}