{"id":3511,"date":"2026-06-05T07:07:00","date_gmt":"2026-06-04T23:07:00","guid":{"rendered":"https:\/\/blog.laoyulaoyu.top\/?p=3511"},"modified":"2026-05-31T23:35:38","modified_gmt":"2026-05-31T15:35:38","slug":"%e4%bb%8e%e7%ba%bf%e6%80%a7%e5%9b%9e%e5%bd%92%e5%88%b0%e6%b7%b1%e5%ba%a6%e6%99%ba%e8%83%bd%e4%bb%a3%e7%90%86%ef%bc%9a%e6%8b%86%e8%a7%a3%e9%87%8f%e5%8c%96%e9%87%91%e8%9e%8d%e9%87%8c%e7%9a%84%e6%9c%ba","status":"publish","type":"post","link":"https:\/\/blog.laoyulaoyu.top\/index.php\/2026\/06\/05\/%e4%bb%8e%e7%ba%bf%e6%80%a7%e5%9b%9e%e5%bd%92%e5%88%b0%e6%b7%b1%e5%ba%a6%e6%99%ba%e8%83%bd%e4%bb%a3%e7%90%86%ef%bc%9a%e6%8b%86%e8%a7%a3%e9%87%8f%e5%8c%96%e9%87%91%e8%9e%8d%e9%87%8c%e7%9a%84%e6%9c%ba\/","title":{"rendered":"\u4ece\u7ebf\u6027\u56de\u5f52\u5230\u6df1\u5ea6\u667a\u80fd\u4ee3\u7406\uff1a\u62c6\u89e3\u91cf\u5316\u91d1\u878d\u91cc\u7684\u673a\u5668\u5b66\u4e60\u5168\u94fe\u6761"},"content":{"rendered":"\n

\u4f5c \u8005\uff1a\u8001\u4f59\u635e\u9c7c<\/a><\/p>\n\n\n\n

\u539f\u521b\u4e0d\u6613\uff0c\u8f6c\u8f7d\u8bf7\u6807\u660e\u51fa\u5904\u53ca\u539f\u4f5c\u8005\u3002<\/mark><\/strong><\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n
\n
\u5199\u5728\u524d\u9762\u7684\u8bdd\uff1a<\/strong>\u5982\u679c\u4f60\u8fd8\u5728\u7528\u5747\u7ebf\u548cMACD\u505a\u51b3\u7b56\uff0c\u90a3\u8fd9\u7bc7\u6587\u7ae0\u4f1a\u6253\u5f00\u4e00\u6247\u65b0\u5927\u95e8\u3002\u4eca\u5929\uff0c\u6211\u628a\u673a\u5668\u5b66\u4e60\u5728\u91cf\u5316\u91d1\u878d\u4e2d\u7684\u5e94\u7528\uff0c\u4ece\u6700\u57fa\u7840\u7684\u7ebf\u6027\u56de\u5f52\u4e00\u8def\u8bb2\u5230\u6700\u524d\u6cbf\u7684\u6df1\u5ea6\u5f3a\u5316\u5b66\u4e60\uff0c\u5168\u90e8\u63b0\u5f00\u63c9\u788e\u4e86\u8bb2\u7ed9\u4f60\u542c<\/mark>\u3002\u4e0d\u6574\u865a\u7684\uff0c\u5168\u662f\u5e72\u8d27\u3002<\/pre>\n<\/blockquote>\n\n\n\n
\u6700\u8fd1\u8ddf\u51e0\u4e2a\u505a\u91cf\u5316\u7684\u670b\u53cb\u804a\u5929\uff0c\u53d1\u73b0\u4e00\u4e2a\u6709\u610f\u601d\u7684\u73b0\u8c61\uff1a\u4ee5\u524d\u5927\u5bb6\u804a\u7684\u662f\u5747\u7ebf\u91d1\u53c9\u3001\u5e03\u6797\u5e26\u7a81\u7834\uff0c\u73b0\u5728\u5f00\u53e3\u95ed\u53e3\u90fd\u662fXGBoost\u3001Transformer\u3001\u6ce8\u610f\u529b\u673a\u5236\u3002<\/strong><\/p>\n\n\n\n
\u8fd9\u4e0d\u662f\u88c5\u903c\uff0c\u800c\u662f\u6574\u4e2a\u884c\u4e1a\u771f\u7684\u5728\u53d1\u751f\u4e00\u573a\u9759\u6084\u6084\u7684\u9769\u547d\u3002<\/p>\n\n\n\n
\u8fc7\u53bb\u5341\u5e74\uff0c\u534e\u5c14\u8857\u548c\u5bf9\u51b2\u57fa\u91d1\u5df2\u7ecf\u628a\u673a\u5668\u5b66\u4e60\u4ece\u5b9e\u9a8c\u5ba4\u642c\u5230\u4e86\u4ea4\u6613\u53f0\u524d\u3002\u6839\u636e\u9ea6\u80af\u95212024\u5e74\u7684\u62a5\u544a\uff0c\u5168\u7403\u5934\u90e8\u5bf9\u51b2\u57fa\u91d1\u4e2d\u8d85\u8fc770%<\/strong>\u5df2\u7ecf\u5728\u751f\u4ea7\u73af\u5883\u4e2d\u90e8\u7f72\u4e86\u67d0\u79cd\u5f62\u5f0f\u7684AI\/ML\u7cfb\u7edf\u3002\u800c\u56fd\u5185\u7684\u91cf\u5316\u5708\uff0c\u4e5f\u5728\u6700\u8fd1\u4e24\u4e09\u5e74\u8fce\u6765\u4e86\u7206\u53d1\u5f0f\u589e\u957f\u3002<\/p>\n\n\n\n
\u4f46\u95ee\u9898\u662f\uff1a\u8fd9\u4e9b\u7b97\u6cd5\u5230\u5e95\u662f\u600e\u4e48\u5de5\u4f5c\u7684\uff1f\u5b83\u4eec\u51ed\u4ec0\u4e48\u80fd\u6bd4\u4f20\u7edf\u65b9\u6cd5\u66f4\u6709\u6548\uff1f\u53c8\u6709\u54ea\u4e9b\u5751\u662f\u65b0\u624b\u4e00\u5b9a\u4f1a\u8e29\u7684\uff1f<\/strong><\/p>\n\n\n\n
\u4eca\u5929\u8fd9\u7bc7\u957f\u6587\uff0c\u6211\u5c31\u6309\u7167\u7531\u6d45\u5165\u6df1\u7684\u987a\u5e8f\uff0c\u628a\u673a\u5668\u5b66\u4e60\u5728\u91d1\u878d\u9886\u57df\u7684\u5b8c\u6574\u77e5\u8bc6\u4f53\u7cfb\u68b3\u7406\u4e00\u904d\u3002\u4e0d\u7ba1\u4f60\u662f\u521a\u5165\u95e8\u7684\u5c0f\u767d\uff0c\u8fd8\u662f\u6709\u4e00\u5b9a\u57fa\u7840\u7684\u4ece\u4e1a\u8005\uff0c\u4e5f\u8bb8\u90fd\u80fd\u4ece\u4e2d\u627e\u5230\u6709\u4ef7\u503c\u7684\u5185\u5bb9\u3002<\/p>\n\n\n\n
\u6ce1\u597d\u5496\u5561\uff0c\u6211\u4eec\u5f00\u59cb\u3002<\/p>\n\n\n\n
\u7b2c\u4e00\u6b65\uff1a\u76d1\u7763\u5b66\u4e60\u6253\u5730\u57fa\uff0c\u4e00\u5207\u4ece\u8fd9\u91cc\u5f00\u59cb<\/strong><\/h2>\n\n\n\n
\u5728\u91d1\u878d\u9886\u57df\uff0c\u6211\u4eec\u624b\u91cc\u6700\u591a\u7684\u4e1c\u897f\u662f\u4ec0\u4e48\uff1f\u5386\u53f2\u6570\u636e\u3002<\/strong>\u80a1\u4ef7\u8d70\u52bf\u3001\u8d22\u52a1\u62a5\u8868\u3001\u5b8f\u89c2\u7ecf\u6d4e\u6307\u6807\u2026\u2026\u8fd9\u4e9b\u6570\u636e\u90fd\u6709\u4e00\u4e2a\u5171\u540c\u7279\u70b9\uff1a\u6211\u4eec\u77e5\u9053\u8fc7\u53bb\u53d1\u751f\u4e86\u4ec0\u4e48<\/strong>\u3002<\/p>\n\n\n\n
\u5f53\u6211\u4eec\u7528\u8fd9\u4e9b”\u6709\u7b54\u6848”\u7684\u6570\u636e\u53bb\u8bad\u7ec3\u6a21\u578b\u65f6\uff0c\u8fd9\u5c31\u53eb\u76d1\u7763\u5b66\u4e60\uff08Supervised Learning<\/em>\uff09<\/strong>\u3002\u5b83\u662f\u6240\u6709\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u4e2d\u6700\u57fa\u7840\u3001\u4e5f\u662f\u6700\u6210\u719f\u7684\u4e00\u7c7b\u3002<\/p>\n\n\n\n
1. \u7ebf\u6027\u56de\u5f52\u4e0e\u8fc7\u62df\u5408\u9677\u9631<\/strong><\/h3>\n\n\n\n
\u4efb\u4f55\u673a\u5668\u5b66\u4e60\u7684\u65c5\u7a0b\uff0c\u51e0\u4e4e\u90fd\u4ece\u7ebf\u6027\u56de\u5f52\uff08Linear Regression\uff09<\/strong>\u5f00\u59cb\u3002<\/p>\n\n\n\n
\u4e3e\u4e2a\u5177\u4f53\u7684\u4f8b\u5b50\uff1a\u5047\u8bbe\u4f60\u60f3\u505a\u4e00\u4e2a\u65f6\u95f4\u5e8f\u5217\u52a8\u91cf\u7b56\u7565\uff0c\u60f3\u7528\u8fc7\u53bb5\u5929\u7684\u6536\u76ca\u7387\u6765\u9884\u6d4b\u660e\u5929\u7684\u6536\u76ca\u7387\u3002\u7ebf\u6027\u56de\u5f52\u505a\u7684\u4e8b\u60c5\u5f88\u7b80\u5355\uff0c\u5b83\u8bd5\u56fe\u5728\u8fd9\u4e9b\u6570\u636e\u70b9\u4e4b\u95f4\u753b\u4e00\u6761\u76f4\u7ebf\uff0c\u627e\u5230\u8f93\u5165\u548c\u8f93\u51fa\u4e4b\u95f4\u7684\u5173\u7cfb\u3002<\/p>\n\n\n\n
\u542c\u8d77\u6765\u5f88\u7f8e\u597d\uff0c\u4f46\u8fd9\u91cc\u6709\u4e00\u4e2a\u5de8\u5927\u7684\u5751\uff1a\u8fc7\u62df\u5408\uff08Overfitting\uff09<\/strong>\u3002<\/p>\n\n\n\n
\u26a0\ufe0f \u8001\u4f59\u63d0\u9192\uff1a\u8fc7\u62df\u5408\u662f\u91cf\u5316\u65b0\u624b\u6700\u5927\u7684\u654c\u4eba\uff01<\/strong><\/p>\n\n\n\n
\u91d1\u878d\u6570\u636e\u7684\u566a\u58f0\u6781\u5927\u3002\u5982\u679c\u4f60\u7684\u6a21\u578b\u592a\u7b80\u5355\uff0c\u5b83\u4f1a\u6f0f\u6389\u771f\u6b63\u7684\u89c4\u5f8b\uff08\u6b20\u62df\u5408\uff09\uff1b\u5982\u679c\u4f60\u8ba9\u6a21\u578b\u53d8\u5f97\u6781\u5176\u590d\u6742\uff08\u6bd4\u5982\u7528\u9ad8\u9636\u591a\u9879\u5f0f\uff09\uff0c\u5b83\u4f1a\u8bb0\u4f4f\u6bcf\u4e00\u4e2a\u6570\u636e\u70b9\u7684\u968f\u673a\u6ce2\u52a8\uff0c\u800c\u4e0d\u662f\u5b66\u5230\u771f\u6b63\u7684\u6a21\u5f0f\u3002\u7ed3\u679c\u5c31\u662f\uff1a\u56de\u6d4b\u770b\u8d77\u6765\u50cf\u5929\u624d\uff0c\u5b9e\u76d8\u4e00\u4e0a\u5c31\u4e8f\u94b1\u3002<\/p>\n\n\n\n
\u6211\u89c1\u8fc7\u592a\u591a\u8fd9\u6837\u7684\u6848\u4f8b\uff1a\u56de\u6d4b\u5e74\u5316\u6536\u76ca80%\uff0c\u590f\u666e\u6bd4\u73873.5\uff0c\u4e00\u5230\u5b9e\u76d8\u5c31\u8fde\u7eed\u4e8f\u635f\u3002\u539f\u56e0\u5f80\u5f80\u4e0d\u662f\u7b56\u7565\u903b\u8f91\u6709\u95ee\u9898\uff0c\u800c\u662f\u8fc7\u62df\u5408\u5728\u4f5c\u795f\u3002<\/p>\n\n\n\n
2. \u6b63\u5219\u5316\uff1a\u7ed9\u6a21\u578b\u8e29\u5239\u8f66<\/strong><\/h3>\n\n\n\n
\u600e\u4e48\u89e3\u51b3\u8fc7\u62df\u5408\uff1f\u91cf\u5316\u5708\u7684\u6807\u51c6\u7b54\u6848\u662f\uff1a\u6b63\u5219\u5316\uff08Regularization\uff09<\/strong>\u3002<\/p>\n\n\n\n
\u4f60\u53ef\u4ee5\u628a\u5b83\u7406\u89e3\u4e3a\u7ed9\u5b66\u4e60\u8fc7\u7a0b\u88c5\u4e86\u4e00\u4e2a”\u5239\u8f66\u88c5\u7f6e”\u3002\u5e38\u7528\u7684\u6b63\u5219\u5316\u65b9\u6cd5\u6709\u4e09\u79cd\uff1a<\/p>\n\n\n\n
\u65b9\u6cd5\u540d\u79f0<\/strong><\/th>\u6838\u5fc3\u601d\u60f3<\/strong><\/th>\u9002\u7528\u573a\u666f<\/strong><\/th><\/tr><\/thead>
Ridge\uff08L2\u6b63\u5219\u5316\uff09<\/strong><\/td>\u60e9\u7f5a\u7cfb\u6570\u7684\u5e73\u65b9\u548c\uff0c\u8ba9\u6240\u6709\u7cfb\u6570\u90fd\u53d8\u5c0f\u4f46\u4e0d\u5f52\u96f6<\/td>\u7279\u5f81\u4e4b\u95f4\u5b58\u5728\u591a\u91cd\u5171\u7ebf\u6027\u65f6<\/td><\/tr>
Lasso\uff08L1\u6b63\u5219\u5316\uff09<\/strong><\/td>\u60e9\u7f5a\u7cfb\u6570\u7684\u7edd\u5bf9\u503c\u4e4b\u548c\uff0c\u80fd\u628a\u4e0d\u91cd\u8981\u7279\u5f81\u7684\u7cfb\u6570\u538b\u7f29\u5230\u96f6<\/td>\u9700\u8981\u505a\u7279\u5f81\u9009\u62e9\u65f6<\/td><\/tr>
ElasticNet<\/strong><\/td>\u7ed3\u5408L1\u548cL2\u7684\u4f18\u70b9\uff0c\u540c\u65f6\u5b9e\u73b0\u6536\u7f29\u548c\u9009\u62e9<\/td>\u7279\u5f81\u591a\u4e14\u76f8\u5173\u6027\u9ad8\u65f6<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
\ud83d\udca1 \u4e00\u53e5\u8bdd\u7406\u89e3\uff1a<\/strong>Ridge\u50cf\u662f\u4e00\u4e2a\u6e29\u548c\u7684\u8001\u5e08\uff0c\u8ba9\u6bcf\u4e2a\u5b66\u751f\u90fd\u5c11\u5199\u4e00\u70b9\u4f5c\u4e1a\uff1bLasso\u5219\u50cf\u4e00\u4e2a\u4e25\u683c\u7684\u8001\u5e08\uff0c\u76f4\u63a5\u544a\u8bc9\u67d0\u4e9b\u5b66\u751f”\u4f60\u4e0d\u7528\u5199\u4e86”\u3002ElasticNet\u5219\u662f\u4e24\u8005\u7684\u5e73\u8861\u3002<\/p>\n\n\n\n
\u7b2c\u4e8c\u6b65\uff1a\u5206\u7c7b\u95ee\u9898\uff0c\u65b9\u5411\u6bd4\u5e45\u5ea6\u66f4\u91cd\u8981<\/strong><\/h2>\n\n\n\n
\u5728\u5b9e\u9645\u4ea4\u6613\u4e2d\uff0c\u9884\u6d4b\u4e00\u53ea\u80a1\u7968\u5177\u4f53\u6da8\u591a\u5c11<\/strong>\u662f\u975e\u5e38\u56f0\u96be\u7684\u3002\u5f88\u591a\u65f6\u5019\uff0c\u6211\u4eec\u53ea\u9700\u8981\u56de\u7b54\u4e00\u4e2a\u66f4\u7b80\u5355\u7684\u95ee\u9898\uff1a\u660e\u5929\u662f\u6da8\u8fd8\u662f\u8dcc\uff1f<\/strong><\/p>\n\n\n\n
\u8fd9\u5c31\u662f\u4ece\u56de\u5f52\u95ee\u9898<\/strong>\u8f6c\u5411\u5206\u7c7b\u95ee\u9898\uff08Classification<\/em>\uff09<\/strong>\u7684\u6838\u5fc3\u52a8\u673a\u3002<\/p>\n\n\n\n
1. \u903b\u8f91\u56de\u5f52\u4e0e\u6734\u7d20\u8d1d\u53f6\u65af<\/strong><\/h3>\n\n\n\n
\u903b\u8f91\u56de\u5f52\uff08Logistic Regression<\/em>\uff09<\/strong>\u867d\u7136\u540d\u5b57\u91cc\u5e26\u7740”\u56de\u5f52”\uff0c\u4f46\u5b83\u5b9e\u9645\u4e0a\u662f\u4e00\u4e2a\u5206\u7c7b\u7b97\u6cd5\u3002\u5b83\u901a\u8fc7\u4e00\u4e2aS\u5f62\u66f2\u7ebf\uff08sigmoid\u51fd\u6570\uff09\uff0c\u628a\u8f93\u51fa\u6620\u5c04\u52300\u52301\u4e4b\u95f4\u7684\u6982\u7387\u503c\u3002\u6bd4\u5982\uff1a”\u5e02\u573a\u660e\u5929\u4e0a\u6da8\u7684\u6982\u7387\u662f65%”\u3002<\/p>\n\n\n\n
\u6734\u7d20\u8d1d\u53f6\u65af\uff08Naive Bayes<\/em>\uff09<\/strong>\u5219\u662f\u53e6\u4e00\u79cd\u601d\u8def\uff0c\u5b83\u57fa\u4e8e\u8d1d\u53f6\u65af\u5b9a\u7406\uff0c\u968f\u7740\u65b0\u8bc1\u636e\u7684\u4e0d\u65ad\u51fa\u73b0\u6765\u66f4\u65b0\u6211\u4eec\u7684\u4fe1\u5ff5\u3002\u6bd4\u5982\u5f53\u4e00\u4efd\u8d85\u9884\u671f\u7684\u8d22\u62a5\u53d1\u5e03\u540e\uff0c\u7cfb\u7edf\u4f1a\u81ea\u52a8\u8c03\u6574\u5bf9\u8fd9\u53ea\u80a1\u7968\u4e0a\u6da8\u6982\u7387\u7684\u5224\u65ad\u3002<\/p>\n\n\n\n
\ud83d\udcca \u91cd\u8981\u63d0\u9192\uff1a\u4e0d\u8981\u8ff7\u4fe1\u51c6\u786e\u7387\uff01<\/strong><\/p>\n\n\n\n
\u5982\u679c\u5e02\u573a\u5929\u7136\u670955%\u7684\u65f6\u95f4\u5728\u4e0a\u6da8\uff0c\u4e00\u4e2a\u6c38\u8fdc\u731c”\u6da8”\u7684\u50bb\u74dc\u6a21\u578b\u4e5f\u80fd\u62ff\u523055%\u7684\u51c6\u786e\u7387\uff0c\u4f46\u5b83\u6ca1\u6709\u4efb\u4f55\u5b9e\u9645\u4ef7\u503c\u3002<\/p>\n\n\n\n
\u4e1a\u5185\u771f\u6b63\u4f7f\u7528\u7684\u8bc4\u4f30\u6307\u6807\u662f ROC\u66f2\u7ebf<\/strong> \u548c AUC\u503c\uff08\u66f2\u7ebf\u4e0b\u9762\u79ef\uff09<\/strong>\uff0c\u5b83\u4eec\u8861\u91cf\u7684\u662f\u6a21\u578b\u533a\u5206”\u8d62\u5bb6”\u548c”\u8f93\u5bb6”\u7684\u771f\u5b9e\u80fd\u529b\u3002<\/p>\n\n\n\n
2. \u652f\u6301\u5411\u91cf\u673a\uff08SVM\uff09\uff1a\u753b\u4e00\u6761\u6700\u5bbd\u7684\u5206\u754c\u7ebf<\/strong><\/h3>\n\n\n\n
\u5f53”\u6da8”\u548c”\u8dcc”\u7684\u5206\u754c\u7ebf\u4e0d\u662f\u4e00\u6761\u76f4\u7ebf\u65f6\u600e\u4e48\u529e\uff1f\u8fd9\u65f6\u5019\u5c31\u8f6e\u5230\u652f\u6301\u5411\u91cf\u673a\uff08Support Vector Machine, SVM<\/em>\uff09<\/strong>\u767b\u573a\u4e86\u3002<\/p>\n\n\n\n
SVM\u7684\u6838\u5fc3\u601d\u60f3\u5f88\u4f18\u96c5\uff1a\u5b83\u8bd5\u56fe\u627e\u5230\u4e00\u4e2a\u8d85\u5e73\u9762\uff08hyperplane\uff09<\/em><\/strong>\uff0c\u4f7f\u5f97\u4e0d\u540c\u7c7b\u522b\u4e4b\u95f4\u7684\u95f4\u9694\uff08margin<\/em>\uff09\u5c3d\u53ef\u80fd\u5927\u3002\u800c\u4e14\u901a\u8fc7\u6838\u6280\u5de7\uff08kernel trick<\/em>\uff09\uff0c\u5b83\u8fd8\u80fd\u5904\u7406\u975e\u7ebf\u6027\u53ef\u5206\u7684\u60c5\u51b5\u3002<\/p>\n\n\n\n
\u4e0d\u8fc7\u8bf4\u5b9e\u8bdd\uff0c\u5728\u5f53\u4eca\u7684\u91cf\u5316\u5b9e\u8df5\u4e2d\uff0cSVM\u7684\u4f7f\u7528\u9891\u7387\u5df2\u7ecf\u4e0d\u5982\u4ece\u524d\u4e86\u3002\u539f\u56e0\u5f88\u7b80\u5355\uff0c\u540e\u9762\u8981\u8bb2\u7684\u6811\u6a21\u578b\u548c\u96c6\u6210\u65b9\u6cd5\uff0c\u5728\u5927\u591a\u6570\u573a\u666f\u4e0b\u6548\u679c\u66f4\u597d\u3002<\/p>\n\n\n\n
\u7b2c\u4e09\u6b65\uff1a\u65e0\u76d1\u7763\u5b66\u4e60\uff0c\u8ba9\u6570\u636e\u81ea\u5df1\u8bf4\u8bdd<\/strong><\/h2>\n\n\n\n
\u524d\u9762\u8bf4\u7684\u90fd\u662f\u6709\u6807\u7b7e\u7684\u5b66\u4e60\uff0c\u6211\u4eec\u77e5\u9053\u6bcf\u6761\u6570\u636e\u5bf9\u5e94\u7684\u7b54\u6848\u662f\u4ec0\u4e48\u3002\u4f46\u5982\u679c\u6ca1\u6709\u6807\u7b7e\u5462\uff1f<\/strong><\/p>\n\n\n\n
\u5047\u8bbe\u4f60\u624b\u4e0a\u6709\u51e0\u5343\u53ea\u80a1\u7968\uff0c\u60f3\u8981\u627e\u51fa\u9690\u85cf\u7684\u6a21\u5f0f\u548c\u7ed3\u6784\u3002\u8fd9\u65f6\u5019\u5c31\u9700\u8981\u65e0\u76d1\u7763\u5b66\u4e60\uff08Unsupervised Learning<\/em>\uff09<\/strong>\u4e86\u3002<\/p>\n\n\n\n
1. \u805a\u7c7b\u5206\u6790\uff1a\u6784\u5efa\u771f\u6b63\u5206\u6563\u7684\u7ec4\u5408<\/strong><\/h3>\n\n\n\n
\u4f20\u7edf\u7684\u884c\u4e1a\u5206\u7c7b\uff08\u79d1\u6280\u3001\u80fd\u6e90\u3001\u533b\u7597\uff09\u770b\u4f3c\u5408\u7406\uff0c\u4f46\u6709\u65f6\u5019\u6570\u636e\u4f1a\u544a\u8bc9\u6211\u4eec\u4e0d\u540c\u7684\u6545\u4e8b\u3002<\/p>\n\n\n\n
\u4f7f\u7528K-Means\u805a\u7c7b<\/strong>\u6216\u5c42\u6b21\u805a\u7c7b\uff08Hierarchical Clustering<\/em>\uff09<\/strong>\uff0c\u6211\u4eec\u53ef\u4ee5\u5b8c\u5168\u57fa\u4e8e\u7edf\u8ba1\u884c\u4e3a\u6765\u5206\u7ec4\u8d44\u4ea7\uff0c\u8003\u8651\u98ce\u9669\u3001\u6536\u76ca\u7387\u3001\u6ce2\u52a8\u7387\u3001Beta\u7b49\u6307\u6807\u3002\u6a21\u578b\u53ef\u80fd\u4f1a\u63ed\u793a\u51fa\uff1a\u67d0\u53ea\u79d1\u6280\u80a1\u7684\u884c\u4e3a\u6a21\u5f0f\u66f4\u50cf\u516c\u7528\u4e8b\u4e1a\u80a1\u3002\u2705 \u5b9e\u6218\u4ef7\u503c\uff1a<\/strong>\u8fd9\u79cd\u57fa\u4e8e\u7edf\u8ba1\u884c\u4e3a\u7684\u805a\u7c7b\uff0c\u80fd\u5e2e\u52a9\u4f60\u6784\u5efa\u6570\u5b66\u610f\u4e49\u4e0a\u771f\u6b63\u5206\u6563\u7684\u6295\u8d44\u7ec4\u5408\uff0c\u800c\u4e0d\u662f\u4ec5\u4ec5\u4f9d\u8d56\u4f20\u7edf\u884c\u4e1a\u5206\u7c7b\u7684\u8868\u9762\u5206\u6563\u3002<\/p>\n\n\n\n
2. \u4e3b\u6210\u5206\u5206\u6790\uff08PCA\uff09\u4e0e\u6536\u76ca\u7387\u66f2\u7ebf<\/strong><\/h3>\n\n\n\n
\u91d1\u878d\u6570\u636e\u96c6\u7ecf\u5e38\u9762\u4e34\u4e00\u4e2a\u53eb“\u7ef4\u5ea6\u8bc5\u5492”\uff08Curse of Dimensionality<\/em>\uff09<\/strong>\u7684\u95ee\u9898\uff0c\u53d8\u91cf\u592a\u591a\u4e14\u9ad8\u5ea6\u76f8\u5173\u3002<\/p>\n\n\n\n
\u4ee5\u7f8e\u56fd\u56fd\u503a\u6536\u76ca\u7387\u66f2\u7ebf\u4e3a\u4f8b\uff1a1\u4e2a\u6708\u30013\u4e2a\u6708\u30011\u5e74\u30015\u5e74\u300110\u5e74\u300130\u5e74\u671f\u56fd\u503a\u6536\u76ca\u7387\u2026\u2026\u8fd9\u4e9b\u53d8\u91cf\u4e4b\u95f4\u9ad8\u5ea6\u76f8\u5173\u3002\u628a\u5b83\u4eec\u5168\u90e8\u653e\u8fdb\u6a21\u578b\uff0c\u4e0d\u4ec5\u8ba1\u7b97\u91cf\u5927\uff0c\u8fd8\u4f1a\u5f15\u5165\u566a\u58f0\u3002<\/p>\n\n\n\n
\u4e3b\u6210\u5206\u5206\u6790\uff08Principal Component Analysis, PCA<\/em>\uff09<\/strong>\u7684\u4f5c\u7528\u5c31\u662f\u628a\u8fd9\u4e9b\u9ad8\u5ea6\u76f8\u5173\u7684\u53d8\u91cf\u538b\u7f29\u6210\u5c11\u6570\u51e0\u4e2a\u72ec\u7acb\u7684”\u4e3b\u6210\u5206”\u3002<\/p>\n\n\n\n
\ud83d\udca1 \u6709\u8da3\u7684\u53d1\u73b0\uff1a<\/strong><\/p>\n\n\n\n
\u5f53\u4f60\u5bf9\u7f8e\u56fd\u56fd\u503a\u6536\u76ca\u7387\u66f2\u7ebf\u505aPCA\u65f6\uff0c\u6570\u5b66\u81ea\u7136\u800c\u7136\u5730\u91cd\u65b0\u53d1\u73b0\u4e86\u503a\u5238\u5e02\u573a\u7684\u4e09\u4e2a\u57fa\u672c\u529b\u91cf\uff1a<\/p>\n\n\n\n
\u2022 \u6c34\u5e73\u56e0\u5b50\uff08Level\uff09<\/strong>\uff1a\u6574\u4f53\u5229\u7387\u6c34\u5e73\u7684\u5e73\u884c\u79fb\u52a8
\u2022 \u659c\u7387\u56e0\u5b50\uff08Slope\uff09<\/strong>\uff1a\u957f\u77ed\u7aef\u5229\u5dee\u7684\u53d8\u5316\uff08\u626d\u66f2\uff09
\u2022 \u66f2\u7387\u56e0\u5b50\uff08Curvature\uff09<\/strong>\uff1a\u4e2d\u671f\u76f8\u5bf9\u4e8e\u4e24\u7aef\u7684\u53d8\u5316\uff08\u8774\u8776\u5f62\u6001\uff09<\/p>\n\n\n\n
\u8fd9\u7edd\u4e0d\u662f\u5de7\u5408\uff0c\u800c\u662f\u8bf4\u660ePCA\u786e\u5b9e\u6355\u6349\u5230\u4e86\u9a71\u52a8\u5e02\u573a\u7684\u5e95\u5c42\u7ecf\u6d4e\u56e0\u7d20\u3002<\/p>\n\n\n\n
\u7b2c\u56db\u6b65\uff1a\u96c6\u6210\u5b66\u4e60\uff0c\u4e09\u4e2a\u81ed\u76ae\u5320\u9876\u4e2a\u8bf8\u845b\u4eae<\/strong><\/h2>\n\n\n\n
\u5355\u4e2a\u51b3\u7b56\u6811\uff08Decision Tree<\/em><\/strong>\uff09\u7684\u597d\u5904\u662f\u53ef\u89e3\u91ca\u6027\u5f3a\uff0c<\/strong>\u5b83\u5c31\u50cf\u4e00\u7cfb\u5217”\u662f\/\u5426”\u95ee\u9898\u7684\u7ec4\u5408\uff08\u6bd4\u5982”\u5e02\u76c8\u7387\u5927\u4e8e15\u5417\uff1f”\uff09\uff0c\u6700\u7ec8\u5f97\u51fa\u7ed3\u8bba\u3002<\/p>\n\n\n\n
\u4f46\u5355\u68f5\u6811\u7684\u95ee\u9898\u4e5f\u5f88\u660e\u663e\uff1a\u5bb9\u6613\u8fc7\u62df\u5408<\/strong>\u3002<\/p>\n\n\n\n
\u89e3\u51b3\u65b9\u6848\uff1f\u96c6\u6210\u65b9\u6cd5\uff08Ensemble Methods\uff09<\/strong>\u628a\u591a\u4e2a\u6a21\u578b\u7ec4\u5408\u8d77\u6765\uff0c\u5f62\u6210\u4e00\u4e2a”\u8d85\u7ea7\u6a21\u578b”\u3002<\/p>\n\n\n\n
\u96c6\u6210\u65b9\u6cd5<\/strong><\/th>\u6838\u5fc3\u601d\u60f3<\/strong><\/th>\u4ee3\u8868\u7b97\u6cd5<\/strong><\/th>\u4f18\u52bf<\/strong><\/th><\/tr><\/thead>
Bagging\uff08\u88c5\u888b\u6cd5\uff09<\/strong><\/td>\u5e76\u884c\u8bad\u7ec3\u591a\u4e2a\u72ec\u7acb\u6a21\u578b\uff0c\u53d6\u6295\u7968\u7ed3\u679c<\/td>Random Forest\uff08\u968f\u673a\u68ee\u6797\uff09<\/td>\u5927\u5e45\u964d\u4f4e\u65b9\u5dee<\/td><\/tr>
Stacking\uff08\u5806\u53e0\u6cd5\uff09<\/strong><\/td>\u8bad\u7ec3\u4e0d\u540c\u7c7b\u578b\u7684\u57fa\u6a21\u578b\uff0c\u518d\u7528\u5143\u6a21\u578b\u878d\u5408<\/td>\u591a\u6a21\u578b\u5806\u53e0<\/td>\u9002\u5e94\u4e0d\u540c\u5e02\u573a\u73af\u5883<\/td><\/tr>
Boosting\uff08\u63d0\u5347\u6cd5\uff09<\/strong><\/td>\u4e32\u884c\u8bad\u7ec3\uff0c\u6bcf\u68f5\u65b0\u6811\u4e13\u95e8\u4fee\u6b63\u524d\u4e00\u68f5\u7684\u9519\u8bef<\/td>XGBoost \/ LightGBM \/ CatBoost<\/td>\u8868\u683c\u6570\u636e\u9886\u57df\u5f53\u524d\u6700\u5f3a<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
XGBoost\uff1a\u8868\u683c\u6570\u636e\u7684\u738b\u8005<\/strong><\/h3>\n\n\n\n
\u8bf4\u5230Boosting\uff0c\u5c31\u4e0d\u5f97\u4e0d\u63d0XGBoost<\/strong>\u3002\u5b83\u662f\u76ee\u524d\u7ed3\u6784\u5316\u6570\u636e\uff08\u8868\u683c\u6570\u636e\uff09\u9886\u57df\u65e0\u53ef\u4e89\u8bae\u7684\u738b\u8005\u3002<\/p>\n\n\n\n
\u5b83\u7684\u539f\u7406\u53ef\u4ee5\u8fd9\u6837\u7406\u89e3\uff1a<\/p>\n\n\n\n
    \n
  • \u7b2c\u4e00\u8f6e<\/strong>\uff1a\u8bad\u7ec3\u4e00\u68f5\u7b80\u5355\u7684\u51b3\u7b56\u6811\u3002<\/li>\n\n\n\n
  • \u7b2c\u4e8c\u8f6e<\/strong>\uff1a\u770b\u770b\u7b2c\u4e00\u68f5\u6811\u54ea\u91cc\u731c\u9519\u4e86\uff08\u6b8b\u5dee\uff09\uff0c\u518d\u8bad\u7ec3\u4e00\u68f5\u6811\u4e13\u95e8\u4fee\u6b63\u8fd9\u4e9b\u9519\u8bef\u3002<\/li>\n\n\n\n
  • \u7b2c\u4e09\u8f6e<\/strong>\uff1a\u7ee7\u7eed\u4fee\u6b63\u524d\u4e24\u68f5\u6811\u7684\u5269\u4f59\u9519\u8bef\u3002<\/li>\n\n\n\n
  • \u2026\u2026\u5982\u6b64\u8fed\u4ee3<\/strong>\uff0c\u76f4\u5230\u8fbe\u5230\u9884\u8bbe\u7684\u8f6e\u6570\u6216\u8bef\u5dee\u8db3\u591f\u5c0f\u3002<\/li>\n<\/ul>\n\n\n\n
    \ud83c\udfc6 \u4e3a\u4ec0\u4e48XGBoost\u5728\u91d1\u878d\u754c\u8fd9\u4e48\u706b\uff1f<\/strong><\/p>\n\n\n\n
    \u6839\u636eKaggle\u7ade\u8d5b\u548c\u5de5\u4e1a\u754c\u7684\u7ecf\u9a8c\uff0cXGBoost\u5728\u4ee5\u4e0b\u91d1\u878d\u4efb\u52a1\u4e2d\u8868\u73b0\u5353\u8d8a\uff1a<\/p>\n\n\n\n
    \u2022 \u4fe1\u7528\u98ce\u9669\u8bc4\u4f30\uff08\u8fdd\u7ea6\u6982\u7387\u9884\u6d4b\uff09
    \u2022 \u6b3a\u8bc8\u68c0\u6d4b\uff08\u4fe1\u7528\u5361\u53cd\u6b3a\u8bc8\uff09
    \u2022 \u5e02\u573a\u65b9\u5411\u9884\u6d4b\uff08\u6da8\/\u8dcc\u5206\u7c7b\uff09
    \u2022 \u5b9a\u4ef7\u6a21\u578b\uff08\u4fdd\u9669\u3001\u884d\u751f\u54c1\u5b9a\u4ef7\uff09<\/p>\n\n\n\n
    \u57282024\u5e74\u7684\u5404\u5927\u91d1\u878dAI\u7ade\u8d5b\u4e2d\uff0cXGBoost\u53ca\u5176\u53d8\u4f53\u4ecd\u7136\u5360\u636e\u7740\u699c\u5355\u524d\u5217\u7684\u4f4d\u7f6e\u3002<\/p>\n\n\n\n
    \u4e0b\u9762\u662f\u4e00\u6bb5XGBoost\u7528\u4e8e\u4fe1\u7528\u98ce\u9669\u5efa\u6a21\u7684\u7b80\u5316\u4ee3\u7801\u793a\u4f8b\uff1a<\/strong><\/p>\n\n\n\n
    # XGBoost \u4fe1\u7528\u98ce\u9669\u5efa\u6a21\u793a\u4f8b<\/code>import xgboost as xgb<\/code>from sklearn.model_selection import train_test_split<\/code>from sklearn.metrics import roc_auc_score<\/code><\/code># \u5047\u8bbe\u5df2\u52a0\u8f7d\u5e76\u9884\u5904\u7406\u597d\u7684\u6570\u636e<\/code># X: \u7279\u5f81\u77e9\u9635 (\u5e74\u9f84, \u6536\u5165, \u8d1f\u503a\u6bd4\u4f8b, \u5386\u53f2\u8fd8\u6b3e\u8bb0\u5f55...)<\/code># y: \u6807\u7b7e (0=\u6b63\u5e38\u8fd8\u6b3e, 1=\u8fdd\u7ea6)<\/code><\/code>X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)<\/code><\/code># \u6784\u5efaXGBoost\u5206\u7c7b\u5668<\/code>model = xgb.XGBClassifier(<\/code>    objective='binary:logistic',  # \u4e8c\u5206\u7c7b\u4efb\u52a1<\/code>    eval_metric='auc',            # \u7528AUC\u4f5c\u4e3a\u8bc4\u4f30\u6307\u6807<\/code>    max_depth=6,                  # \u6811\u7684\u6700\u5927\u6df1\u5ea6<\/code>    learning_rate=0.05,           # \u5b66\u4e60\u7387<\/code>    n_estimators=300,             # \u8fed\u4ee3\u8f6e\u6570<\/code>    scale_pos_weight=10           # \u5904\u7406\u7c7b\u522b\u4e0d\u5e73\u8861<\/code>)<\/code><\/code># \u8bad\u7ec3\u6a21\u578b\uff08\u5e26\u65e9\u505c\u673a\u5236\u9632\u6b62\u8fc7\u62df\u5408\uff09<\/code>model.fit(<\/code>    X_train, y_train,<\/code>    eval_set=[(X_test, y_test)],<\/code>    early_stopping_rounds=20,<\/code>    verbose=False<\/code>)<\/code><\/code># \u9884\u6d4b\u5e76\u8bc4\u4f30<\/code>y_pred_proba = model.predict_proba(X_test)[:, 1]<\/code>print(f\"\u6d4b\u8bd5\u96c6AUC: {roc_auc_score(y_test, y_pred_proba):.4f}\")<\/code><\/code><\/pre>\n\n\n\n
    \u7b2c\u4e94\u6b65\uff1a\u6df1\u5ea6\u5b66\u4e60\uff0c\u8fdb\u5165\u6df1\u6c34\u533a<\/strong><\/h2>\n\n\n\n
    \u5f53\u4f20\u7edf\u673a\u5668\u5b66\u4e60\u65b9\u6cd5\u9047\u5230\u74f6\u9888\u65f6\uff0c\u5c31\u8be5\u6df1\u5ea6\u5b66\u4e60\uff08Deep Learning<\/em>\uff09<\/strong>\u51fa\u573a\u4e86\u3002<\/p>\n\n\n\n
    \u4eba\u5de5\u795e\u7ecf\u7f51\u7edc\uff08Artificial Neural Networks, ANNs<\/em>\uff09<\/strong>\u7684\u7075\u611f\u6765\u6e90\u4e8e\u4eba\u8111\u7684\u7ed3\u6784\u3002\u6570\u636e\u901a\u8fc7\u591a\u5c42”\u795e\u7ecf\u5143”\u4f20\u9012\uff0c\u6bcf\u4e00\u5c42\u90fd\u4f1a\u5e94\u7528\u975e\u7ebf\u6027\u7684\u6fc0\u6d3b\u51fd\u6570\uff08\u5982ReLU\uff09\uff0c\u4ece\u800c\u6355\u6349\u6781\u5176\u590d\u6742\u7684\u5e02\u573a\u52a8\u6001\u3002<\/p>\n\n\n\n
    \u6df1\u5ea6\u5b66\u4e60\u7684\u771f\u6b63\u9b54\u6cd5\uff1a\u4f18\u5316<\/strong><\/h3>\n\n\n\n
    \u795e\u7ecf\u7f51\u7edc\u7684\u5de5\u4f5c\u6d41\u7a0b\u53ef\u4ee5\u6982\u62ec\u4e3a\u4e09\u6b65\u5faa\u73af\uff1a<\/p>\n\n\n\n
      \n
    • \u524d\u5411\u4f20\u64ad<\/strong>\uff1a\u6570\u636e\u4ece\u8f93\u5165\u5c42\u7ecf\u8fc7\u5404\u5c42\u5904\u7406\uff0c\u5f97\u5230\u9884\u6d4b\u7ed3\u679c\u3002<\/li>\n\n\n\n
    • \u635f\u5931\u8ba1\u7b97<\/strong>\uff1a\u5bf9\u6bd4\u9884\u6d4b\u503c\u548c\u771f\u5b9e\u503c\uff0c\u8ba1\u7b97\u8bef\u5dee\uff08Loss Function<\/em>\uff09\u3002<\/li>\n\n\n\n
    • \u53cd\u5411\u4f20\u64ad<\/strong>\uff1a\u5229\u7528\u5fae\u79ef\u5206\uff08\u68af\u5ea6\u4e0b\u964d\uff09\u66f4\u65b0\u6bcf\u4e2a\u795e\u7ecf\u5143\u7684\u6743\u91cd\u53c2\u6570\u3002<\/li>\n<\/ul>\n\n\n\n
      \u73b0\u4ee3\u4f18\u5316\u5668\u5982Adam<\/strong>\uff0c\u5c31\u50cf\u4e00\u4e2a\u806a\u660e\u7684\u767b\u5c71\u8005\u2014\u2014\u5b83\u5728\u8ff7\u96fe\u7b3c\u7f69\u7684\u5c71\u5761\u4e0a\u884c\u8d70\uff0c\u80fd\u52a8\u6001\u8c03\u6574\u6b65\u4f10\u5927\u5c0f\uff0c\u52aa\u529b\u627e\u5230\u5168\u5c40\u6700\u4f4e\u70b9\uff08\u5168\u5c40\u6700\u5c0f\u503c\uff09\uff0c\u800c\u4e0d\u662f\u5361\u5728\u67d0\u4e2a\u5c40\u90e8\u5751\u6d3c\u91cc\uff08\u5c40\u90e8\u6700\u5c0f\u503c\uff09\u3002<\/p>\n\n\n\n
      \"\"<\/figure>\n\n\n\n

      \u7b2c\u516d\u6b65\uff1a\u524d\u6cbf\u5e94\u7528\uff0c\u73b0\u4ee3\u5bf9\u51b2\u57fa\u91d1\u7684\u6b66\u5668<\/strong><\/h2>\n\n\n\n
      \u638c\u63e1\u4e86\u57fa\u7840\u7b97\u6cd5\u4e4b\u540e\uff0c\u6211\u4eec\u6765\u770b\u770b\u9876\u7ea7\u673a\u6784\u6b63\u5728\u7528\u7684\u90a3\u4e9b\u66f4\u9ad8\u7ea7\u7684\u6280\u672f\u3002<\/p>\n\n\n\n
      1. \u81ea\u7136\u8bed\u8a00\u5904\u7406\uff08NLP\uff09\uff1a\u8bfb\u61c2\u6587\u5b57\u91cc\u7684\u4fe1\u53f7<\/strong><\/h3>\n\n\n\n
      \u6570\u5b57\u53ea\u80fd\u544a\u8bc9\u4f60\u4e00\u534a\u7684\u6545\u4e8b\u3002\u53e6\u4e00\u534a\u85cf\u5728\u6587\u5b57\u91cc\u3002<\/p>\n\n\n\n
      \u91cf\u5316\u56e2\u961f\u73b0\u5728\u5927\u91cf\u4f7f\u7528NLP\u6a21\u578b\uff08\u5982Transformer\u67b6\u6784\u3001FinBERT\uff09\u6765\u5b9e\u65f6”\u9605\u8bfb”\u4ee5\u4e0b\u5185\u5bb9\uff1a<\/p>\n\n\n\n
        \n
      • SEC\u76d1\u7ba1\u6587\u4ef6\uff08\u5e74\u62a5\u3001\u5b63\u62a5\u3001\u516c\u544a\uff09\uff1b<\/li>\n\n\n\n
      • \u8d22\u62a5\u7535\u8bdd\u4f1a\u8bae\u7684\u6587\u5b57\u8bb0\u5f55\uff1b<\/li>\n\n\n\n
      • \u5b9e\u65f6\u65b0\u95fb\u63a8\u9001\u548c\u793e\u4ea4\u5a92\u4f53\u60c5\u7eea\uff1b<\/li>\n\n\n\n
      • \u592e\u884c\u5b98\u5458\u7684\u8bb2\u8bdd\u7a3f\uff1b<\/li>\n<\/ul>\n\n\n\n
        \u901a\u8fc7\u63d0\u53d6\u60c5\u7eea\u5206\u6570\uff08Sentiment Score<\/em>\uff09<\/strong>\u6bd4\u5982\u8861\u91cfCEO\u53d1\u8a00\u4e2d\u4e50\u89c2\u8bcd\u6c47\u4e0e\u60b2\u89c2\u8bcd\u6c47\u7684\u6bd4\u4f8b\uff0c\u7cfb\u7edf\u80fd\u591f\u5728\u4eba\u7c7b\u5206\u6790\u5e08\u8bfb\u5b8c\u6807\u9898\u4e4b\u524d\u5c31\u5b8c\u6210\u4fe1\u606f\u6d88\u5316\u548c\u51b3\u7b56\u3002<\/p>\n\n\n\n
        \u2705 \u5b9e\u9645\u6848\u4f8b\uff1a<\/strong>2024\u5e74\u7684\u4e00\u9879\u7814\u7a76\u663e\u793a\uff0c\u57fa\u4e8eFinBERT\u7684\u7f8e\u8054\u50a8\u8bae\u606f\u4f1a\u8bae\u7eaa\u8981\u60c5\u7eea\u5206\u6790\uff0c\u80fd\u591f\u63d0\u524d\u9884\u6d4b\u7ea660%\u7684\u8d27\u5e01\u653f\u7b56\u8d70\u5411\u53d8\u5316\uff0c\u4e3a\u56fa\u5b9a\u6536\u76ca\u4ea4\u6613\u63d0\u4f9b\u4e86\u663e\u8457\u7684alpha\u6765\u6e90\u3002<\/p>\n\n\n\n
        2. \u5f3a\u5316\u5b66\u4e60\uff08RL\uff09\uff1a\u4ece\u8bd5\u9519\u4e2d\u5b66\u4f1a\u6700\u4f18\u7b56\u7565<\/strong><\/h3>\n\n\n\n
        \u524d\u9762\u7684\u65b9\u6cd5\u90fd\u662f\u5728”\u9884\u6d4b\u4e0b\u4e00\u4e2a\u4ef7\u683c”\u3002\u800c\u5f3a\u5316\u5b66\u4e60\uff08Reinforcement Learning<\/em>\uff09<\/strong>\u8d70\u4e86\u4e00\u6761\u5b8c\u5168\u4e0d\u540c\u7684\u8def\uff0c\u5b83\u4e0d\u9884\u6d4b\u4ef7\u683c\uff0c\u800c\u662f\u5b66\u4e60\u5982\u4f55\u505a\u51b3\u7b56<\/strong>\u3002<\/p>\n\n\n\n
        \u4f60\u53ef\u4ee5\u628aRL\u4ea4\u6613\u667a\u80fd\u4f53\u60f3\u8c61\u6210\u4e00\u4e2a\u7535\u5b50\u6e38\u620f\u89d2\u8272\uff1a<\/p>\n\n\n\n
          \n
        • \u73af\u5883<\/strong>\uff1a\u80a1\u7968\u5e02\u573a\uff1b<\/li>\n\n\n\n
        • \u52a8\u4f5c<\/strong>\uff1a\u6301\u6709 \/ \u64cd\u4f5c\uff08\u6839\u636e\u7b56\u7565\u89c4\u5219\u6267\u884c\uff09\uff1b<\/li>\n\n\n\n
        • \u5956\u52b1<\/strong>\uff1a\u76c8\u5229\uff08\u6b63\u5956\u52b1\uff09\u6216\u4e8f\u635f\uff08\u8d1f\u5956\u52b1\uff09\u3002<\/li>\n<\/ul>\n\n\n\n
          \u7ecf\u8fc7\u6570\u767e\u4e07\u6b21\u7684\u6a21\u62df\u8bad\u7ec3\uff0cQ-Learning\u7b49\u7b97\u6cd5\u80fd\u8ba9\u667a\u80fd\u4f53\u9010\u6b65\u5b66\u4f1a\u5728\u4e0d\u540c\u5e02\u573a\u72b6\u6001\u4e0b\u91c7\u53d6\u6700\u4f18\u7684\u884c\u52a8\u5e8f\u5217\uff0c\u4ee5\u6700\u5927\u5316\u957f\u671f\u8d22\u5bcc\u79ef\u7d2f\u3002\u26a0\ufe0f \u6ce8\u610f\uff1a<\/strong>\u5f3a\u5316\u5b66\u4e60\u5728\u91d1\u878d\u4e2d\u7684\u5e94\u7528\u4ecd\u5904\u4e8e\u63a2\u7d22\u9636\u6bb5\u3002\u6a21\u62df\u73af\u5883\u548c\u771f\u5b9e\u5e02\u573a\u4e4b\u95f4\u5b58\u5728\u5de8\u5927\u5dee\u8ddd\uff08\u6240\u8c13\u7684”\u6a21\u62df-\u73b0\u5b9e\u9e3f\u6c9f”\uff09\uff0c\u76f4\u63a5\u5c06RL\u7b56\u7565\u90e8\u7f72\u5230\u5b9e\u76d8\u9700\u8981\u6781\u5176\u8c28\u614e\u7684\u9a8c\u8bc1\u8fc7\u7a0b\u3002<\/p>\n\n\n\n
          3. \u53ef\u89e3\u91ca\u6027AI\uff08XAI\uff09\uff1a\u6253\u5f00\u9ed1\u76d2\u5b50<\/strong><\/h3>\n\n\n\n
          \u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u7684\u4e00\u4e2a\u81f4\u547d\u5f31\u70b9\u662f\uff1a\u9ed1\u7bb1\u7279\u6027<\/strong>\u3002\u5f53\u4e00\u4e2aML\u7cfb\u7edf\u62d2\u7edd\u4e86\u4e00\u7b14\u8d37\u6b3e\u7533\u8bf7\u65f6\uff0c\u76d1\u7ba1\u673a\u6784\u548c\u5ba2\u6237\u90fd\u6709\u6743\u77e5\u9053\u4e3a\u4ec0\u4e48\u3002<\/p>\n\n\n\n
          SHAP\uff08SHapley Additive exPlanations<\/em>\uff09<\/strong>\u662f\u76ee\u524d\u4e1a\u754c\u6700\u6d41\u884c\u7684\u89e3\u91ca\u5de5\u5177\u4e4b\u4e00\u3002\u5b83\u80fd\u9006\u5411\u5206\u89e3\u6a21\u578b\u7684\u51b3\u7b56\u8fc7\u7a0b\uff0c\u7cbe\u786e\u5730\u544a\u8bc9\u4f60\u662f\u54ea\u4e9b\u7279\u5f81\u5bfc\u81f4\u4e86\u67d0\u4e2a\u7279\u5b9a\u7684\u9884\u6d4b\u7ed3\u679c\u3002<\/p>\n\n\n\n
          \ud83d\udca1 SHAP\u7684\u5b9e\u9645\u4ef7\u503c\uff1a<\/strong><\/p>\n\n\n\n
          \u5047\u8bbe\u94f6\u884c\u62d2\u7edd\u4e86\u4e00\u4e2a\u8d37\u6b3e\u7533\u8bf7\uff0cSHAP\u53ef\u4ee5\u8ba9\u94f6\u884c\u660e\u786e\u5730\u8bf4\uff1a”\u60a8\u7684\u7533\u8bf7\u88ab\u62d2\u7edd\uff0c\u662f\u56e0\u4e3a\u8fd1\u671f\u7684\u5f81\u4fe1\u67e5\u8be2\u6b21\u6570\u8fc7\u591a\u8fd9\u4e00\u8d1f\u9762\u56e0\u7d20\uff0c\u8d85\u8fc7\u4e86\u60a8\u7a33\u5b9a\u6536\u5165\u5e26\u6765\u7684\u6b63\u9762\u5f71\u54cd\u3002”\u8fd9\u79cd\u900f\u660e\u5ea6\u5bf9\u4e8e\u5408\u89c4\u548c\u5ba2\u6237\u4fe1\u4efb\u81f3\u5173\u91cd\u8981\u3002<\/p>\n\n\n\n
          SHAP\u89e3\u91caXGBoost\u6a21\u578b\u7684\u4ee3\u7801\u793a\u4f8b\uff1a<\/strong><\/p>\n\n\n\n
          import shap<\/code>import xgboost<\/code><\/code># \u8bad\u7ec3\u597d\u7684XGBoost\u6a21\u578b<\/code>model = xgboost.train(params, dtrain)<\/code><\/code># \u521b\u5efaSHAP\u89e3\u91ca\u5668<\/code>explainer = shap.TreeExplainer(model)<\/code><\/code># \u8ba1\u7b97SHAP\u503c\uff08\u89e3\u91ca\u5355\u4e2a\u9884\u6d4b\uff09<\/code>shap_values = explainer.shap_values(X_test)<\/code><\/code># \u53ef\u89c6\u5316\u5355\u4e2a\u6837\u672c\u7684\u7279\u5f81\u8d21\u732e<\/code>shap.force_plot(explainer.expected_value, shap_values[0], X_test.iloc[0], matplotlib=True)<\/code><\/code><\/pre>\n\n\n\n
          \u7b2c\u4e03\u6b65\uff1a\u5b9e\u6218\u4e2d\u7684\u62e6\u8def\u864e<\/strong><\/h2>\n\n\n\n
          \u4e86\u89e3\u4e86\u7b97\u6cd5\u539f\u7406\u53ea\u662f\u7b2c\u4e00\u6b65\u3002\u8981\u628a\u8fd9\u4e9b\u65b9\u6cd5\u771f\u6b63\u7528\u5230 messy \u7684\u771f\u5b9e\u4e16\u754c\u4e2d\uff0c\u4f60\u8fd8\u9700\u8981\u8de8\u8fc7\u597d\u51e0\u9053\u574e\u3002<\/p>\n\n\n\n
          1. \u6570\u636e\u4e0d\u5e73\u8861\u95ee\u9898\uff08SMOTE\uff09<\/strong><\/h3>\n\n\n\n
          \u5728\u4fe1\u7528\u98ce\u9669\u5efa\u6a21\u4e2d\uff0c99%\u7684\u4eba\u6309\u65f6\u8fd8\u8d37\uff0c\u53ea\u67091%\u7684\u4eba\u8fdd\u7ea6\u3002\u5982\u679c\u6a21\u578b\u6bcf\u6b21\u90fd\u731c”\u4e0d\u4f1a\u8fdd\u7ea6”\uff0c\u5b83\u4e5f\u80fd\u58f0\u79f099%\u7684\u51c6\u786e\u7387\uff0c\u4f46\u8fd9\u6beb\u65e0\u610f\u4e49\u3002<\/p>\n\n\n\n
          \u4e1a\u754c\u7684\u6807\u51c6\u505a\u6cd5\u4e4b\u4e00\u662f\u4f7f\u7528SMOTE\uff08Synthetic Minority Over-sampling Technique<\/em>\uff09\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u5408\u6210\u4e00\u4e9b\u5408\u7406\u7684”\u5047”\u8fdd\u7ea6\u6837\u672c\uff0c\u5f3a\u5236\u7b97\u6cd5\u53bb\u5b66\u4e60\u8fdd\u7ea6\u6848\u4f8b\u7684\u7279\u5f81\u6a21\u5f0f\u3002<\/strong><\/p>\n\n\n\n
          2. \u8d85\u53c2\u6570\u8c03\u4f18\uff1a\u5bfb\u627e\u6700\u4f73\u914d\u7f6e<\/strong><\/h3>\n\n\n\n
          \u6bcf\u4e2aML\u6a21\u578b\u90fd\u6709\u4e00\u5806\u53eb\u505a\u8d85\u53c2\u6570\uff08Hyperparameters\uff09<\/strong>\u7684\u8bbe\u7f6e\u9879\u2014\u2014\u6bd4\u5982\u6811\u7684\u6df1\u5ea6\u3001\u795e\u7ecf\u7f51\u7edc\u7684\u5b66\u4e60\u7387\u3001\u6b63\u5219\u5316\u7684\u5f3a\u5ea6\u7b49\u7b49\u3002<\/p>\n\n\n\n
          \u8c03\u4f18\u65b9\u6cd5<\/strong><\/th>\u539f\u7406<\/strong><\/th>\u4f18\u70b9<\/strong><\/th>\u7f3a\u70b9<\/strong><\/th><\/tr><\/thead>
          \u7f51\u683c\u641c\u7d22<\/strong><\/td>\u7a77\u4e3e\u6240\u6709\u53c2\u6570\u7ec4\u5408<\/td>\u4fdd\u8bc1\u627e\u5230\u6700\u4f18<\/td>\u6781\u6162\uff0c\u7ef4\u5ea6\u707e\u96be<\/td><\/tr>
          \u968f\u673a\u641c\u7d22<\/strong><\/td>\u968f\u673a\u91c7\u6837\u53c2\u6570\u7a7a\u95f4<\/td>\u6bd4\u7f51\u683c\u641c\u7d22\u5feb<\/td>\u4f9d\u8d56\u8fd0\u6c14<\/td><\/tr>
          \u8d1d\u53f6\u65af\u4f18\u5316<\/strong><\/td>\u5229\u7528\u5386\u53f2\u7ed3\u679c\u6307\u5bfc\u4e0b\u4e00\u6b65\u641c\u7d22<\/td>\u9ad8\u6548\uff0c\u667a\u80fd<\/td>\u5b9e\u73b0\u8f83\u590d\u6742<\/td><\/tr>
          \u6a21\u62df\u9000\u706b<\/strong><\/td>\u6a21\u4eff\u91d1\u5c5e\u9000\u706b\u8fc7\u7a0b\u7684\u542f\u53d1\u5f0f\u641c\u7d22<\/td>\u907f\u514d\u5c40\u90e8\u6700\u4f18<\/td>\u9700\u8c03\u8282\u6e29\u5ea6\u53c2\u6570<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
          \ud83d\udd27 \u8001\u4f59\u7684\u7ecf\u9a8c\u4e4b\u8c08\uff1a<\/strong><\/p>\n\n\n\n
          \u8d1d\u53f6\u65af\u4f18\u5316\u548c\u6a21\u62df\u9000\u706b\u5c31\u50cf\u662f\u73a9”\u6218\u8230\u68cb”\u7684\u9ad8\u624b\uff1a\u5b83\u4eec\u4f1a\u5229\u7528\u4e4b\u524d\u7684”\u547d\u4e2d”\u548c”\u672a\u547d\u4e2d”\u8bb0\u5f55\uff0c\u806a\u660e\u5730\u63a8\u6d4b\u6700\u4f73\u53c2\u6570\u85cf\u5728\u54ea\u91cc\u3002\u8fd9\u6bd4\u76f2\u76ee\u641c\u7d22\u6548\u7387\u9ad8\u51fa\u6570\u500d\uff0c\u5bf9\u4e8e\u9700\u8981\u9891\u7e41\u4f18\u5316\u7684\u5b9e\u76d8\u7b56\u7565\u6765\u8bf4\u5c24\u5176\u91cd\u8981\u3002<\/p>\n\n\n\n
          3. \u907f\u514d\u524d\u89c6\u504f\u5dee\uff08Look-ahead Bias\uff09<\/strong><\/h3>\n\n\n\n
          \u8fd9\u662f\u91cf\u5316\u7814\u7a76\u4e2d\u6700\u5bb9\u6613\u72af\u4e5f\u6700\u81f4\u547d\u7684\u9519\u8bef\u4e4b\u4e00<\/strong>\u3002<\/p>\n\n\n\n
          \u7b80\u5355\u6765\u8bf4\uff0c\u5c31\u662f\u5728\u8bad\u7ec3\u6a21\u578b\u65f6\u4e0d\u5c0f\u5fc3\u4f7f\u7528\u4e86”\u672a\u6765\u624d\u5e94\u8be5\u77e5\u9053”\u7684\u4fe1\u606f\u3002\u6bd4\u5982\u7528\u5f53\u5929\u6536\u76d8\u4ef7\u6765\u9884\u6d4b\u5f53\u5929\u76d8\u4e2d\u8d70\u52bf\uff0c\u6216\u8005\u5728\u8ba1\u7b97\u6280\u672f\u6307\u6807\u65f6\u4f7f\u7528\u4e86\u5305\u542b\u5f53\u671f\u5728\u5185\u7684\u672a\u6765\u6570\u636e\u3002<\/p>\n\n\n\n
          \u26a0\ufe0f \u8840\u6cea\u6559\u8bad\uff1a<\/strong><\/p>\n\n\n\n
          \u6211\u89c1\u8fc7\u65e0\u6570\u56de\u6d4b\u5b8c\u7f8e\u7684\u7b56\u7565\uff0c\u4ed4\u7ec6\u5ba1\u67e5\u540e\u53d1\u73b0\u90fd\u5b58\u5728\u4e0d\u540c\u7a0b\u5ea6\u7684\u524d\u89c6\u504f\u5dee\u3002\u4e00\u65e6\u6d88\u9664\u8fd9\u4e2a\u504f\u5dee\uff0c\u5f88\u591a”\u5723\u676f\u7ea7”\u7b56\u7565\u77ac\u95f4\u53d8\u6210\u4e86\u5e73\u5eb8\u751a\u81f3\u4e8f\u635f\u7684\u7b56\u7565\u3002<\/p>\n\n\n\n
          \u9632\u8303\u539f\u5219\uff1a\u4e25\u683c\u6309\u65f6\u95f4\u987a\u5e8f\u5207\u5206\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\uff0c\u786e\u4fdd\u6a21\u578b\u5728\u4efb\u4f55\u65f6\u523b\u53ea\u80fd\u4f7f\u7528\u8be5\u65f6\u523b\u4e4b\u524d\u7684\u4fe1\u606f\u3002<\/strong><\/p>\n\n\n\n
          \u5199\u5728\u6700\u540e<\/strong><\/h2>\n\n\n\n
          \u770b\u5230\u8fd9\u91cc\uff0c\u4f60\u5e94\u8be5\u5df2\u7ecf\u5bf9\u673a\u5668\u5b66\u4e60\u5728\u91cf\u5316\u91d1\u878d\u4e2d\u7684\u5e94\u7528\u6709\u4e86\u6bd4\u8f83\u5b8c\u6574\u7684\u8ba4\u8bc6\u3002<\/p>\n\n\n\n
          \u6211\u60f3\u5f3a\u8c03\u7684\u662f\uff1aML\u5728\u91d1\u878d\u4e2d\u7684\u5e94\u7528\uff0c\u8fdc\u4e0d\u662f”\u5f80\u5899\u4e0a\u6254\u6570\u5b66\u516c\u5f0f\u770b\u54ea\u4e2a\u80fd\u7c98\u4f4f”\u90a3\u4e48\u7b80\u5355\u3002<\/strong> \u5b83\u662f\u4e00\u6761\u4e25\u8c28\u7684\u3001\u73af\u73af\u76f8\u6263\u7684\u6d41\u6c34\u7ebf\uff0c\u4f60\u9700\u8981\u7cbe\u5fc3\u8bbe\u8ba1\u7279\u5f81\u5de5\u7a0b\uff0c\u9009\u62e9\u80fd\u5728\u504f\u5dee\u548c\u65b9\u5dee\u4e4b\u95f4\u53d6\u5f97\u5e73\u8861\u7684\u6a21\u578b\uff0c\u514b\u670d\u6570\u636e\u4e0d\u5e73\u8861\u95ee\u9898\uff0c\u8fdb\u884c\u667a\u80fd\u7684\u8d85\u53c2\u6570\u4f18\u5316\uff0c\u5e76\u4e14\u5728\u6574\u4e2a\u8fc7\u7a0b\u4e2d\u4e25\u9632\u524d\u89c6\u504f\u5dee\u7b49\u5404\u79cd\u9677\u9631\u3002<\/p>\n\n\n\n
          \u65e0\u8bba\u4f60\u662f\u5728\u7528XGBoost\u9884\u6d4b\u4fe1\u7528\u8fdd\u7ea6\uff0c\u8fd8\u662f\u7528NLP\u89e3\u6790\u7f8e\u8054\u50a8\u4e3b\u5e2d\u7684\u8bb2\u8bdd\uff0c\u4ea6\u6216\u662f\u7528\u6df1\u5ea6\u5b66\u4e60\u5efa\u6a21\u6536\u76ca\u7387\u66f2\u7ebf\uff0c\u4f60\u90fd\u6b63\u5728\u53c2\u4e0e\u4eba\u7c7b\u5386\u53f2\u4e0a\u6700\u7cbe\u5bc6\u7684\u91cf\u5316\u91d1\u878d\u65f6\u4ee3\u3002<\/p>\n\n\n\n
          \u5f53\u7136\uff0c\u5de5\u5177\u672c\u8eab\u4e0d\u662f\u76ee\u7684\u3002\u518d\u5148\u8fdb\u7684\u7b97\u6cd5\uff0c\u4e5f\u9700\u8981\u6b63\u786e\u7684\u6295\u8d44\u7406\u5ff5\u3001\u4e25\u8c28\u7684\u98ce\u63a7\u4f53\u7cfb\u548c\u6301\u7eed\u8fed\u4ee3\u7684\u6267\u884c\u529b\u6765\u914d\u5408\u3002\u6280\u672f\u662f\u6760\u6746\uff0c\u4f46\u652f\u70b9\u59cb\u7ec8\u662f\u4f60\u5bf9\u5e02\u573a\u7684\u7406\u89e3\u3002<\/p>\n\n\n\n
          \u672c\u6587\u8981\u70b9\u56de\u987e<\/strong><\/h2>\n\n\n\n