{"id":2979,"date":"2022-02-20T20:56:05","date_gmt":"2022-02-20T12:56:05","guid":{"rendered":"https:\/\/egonlin.com\/?p=2979"},"modified":"2022-02-20T20:56:31","modified_gmt":"2022-02-20T12:56:31","slug":"%e7%ac%ac%e4%ba%8c%e8%8a%82%ef%bc%9ascikit-learn%e5%ba%93%e4%b9%8b%e7%ba%bf%e6%80%a7%e5%9b%9e%e5%bd%92","status":"publish","type":"post","link":"https:\/\/egonlin.com\/?p=2979","title":{"rendered":"\u7b2c\u4e8c\u8282\uff1ascikit-learn\u5e93\u4e4b\u7ebf\u6027\u56de\u5f52"},"content":{"rendered":"

scikit-learn\u5e93\u4e4b\u7ebf\u6027\u56de\u5f52<\/h1>\n

\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"<\/div><\/p>\n

  \u7531\u4e8escikit-learn\u5e93\u4e2dsclearn.linear_model<\/code>\u63d0\u4f9b\u4e86\u591a\u79cd\u652f\u6301\u7ebf\u6027\u56de\u5f52\u5206\u6790\u7684\u7c7b\uff0c\u672c\u6587\u4e3b\u8981\u603b\u7ed3\u4e00\u4e9b\u5e38\u7528\u7684\u7ebf\u6027\u56de\u5f52\u7684\u7c7b\uff0c\u5e76\u4e14\u7531\u4e8e\u662f\u4ece\u5b98\u65b9\u6587\u6863\u7ffb\u8bd1\u800c\u6765\uff0c\u7ffb\u8bd1\u4f1a\u7565\u6709\u504f\u9887\uff0c\u5982\u679c\u6709\u5174\u8da3\u60f3\u4e86\u89e3\u5176\u4ed6\u7c7b\u7684\u4f7f\u7528\u65b9\u6cd5\u7684\u540c\u5b66\u4e5f\u53ef\u4ee5\u53bbscikit-learn\u5b98\u65b9\u6587\u6863\u67e5\u770bhttps:\/\/scikit-learn.org\/stable\/modules\/classes.html#module-sklearn.linear_model<\/p>\n

  \u5728\u8bb2\u7ebf\u6027\u56de\u5f52\u7406\u8bba\u7684\u65f6\u5019\u8bb2\u5230\u4e86\uff0c\u7ebf\u6027\u56de\u5f52\u7684\u76ee\u7684\u662f\u627e\u5230\u4e00\u4e2a\u7ebf\u6027\u56de\u5f52\u7cfb\u6570\u5411\u91cf$\\omega$\uff0c\u4f7f\u5f97\u8f93\u5165\u7279\u5f81$X$\u548c\u8f93\u51fa\u5411\u91cf$Y$\u4e4b\u95f4\u6709\u4e00\u4e2a
\n$$
\nY = X\\omega
\n$$
\n\u7684\u6620\u5c04\u5173\u7cfb\uff0c\u63a5\u4e0b\u6765\u7684\u7ebf\u6027\u56de\u5f52\u6a21\u578b\u548c\u7ebf\u6027\u56de\u5f52\u6a21\u578b\u7684\u601d\u60f3\u7c7b\u4f3c\u3002\u5047\u8bbe\u4e00\u4e2a\u6570\u636e\u96c6\u6709$m$\u5b9e\u4f8b\uff0c\u6bcf\u4e2a\u5b9e\u4f8b\u6709$n$\u4e2a\u7279\u5f81\uff0c\u5219\u5176\u4e2d$Y$\u7684\u7ef4\u5ea6\u662f$m1$\uff0c$X$\u7684\u7ef4\u5ea6\u662f$m<\/em>n$\uff0c$\\omega$\u7684\u7ef4\u5ea6\u662f$n*1$\u3002<\/p>\n

  \u4f7f\u7528\u7ebf\u6027\u56de\u5f52\u7684\u76ee\u7684\u5c31\u662f\u627e\u5230\u4e00\u4e2a\u5408\u9002\u7684\u7ebf\u6027\u56de\u5f52\u7cfb\u6570$\\omega$\u80fd\u591f\u6700\u5c0f\u5316\u6211\u4eec\u5b9a\u4e49\u7684\u76ee\u6807\u51fd\u6570\uff0c\u53c8\u7531\u4e8e\u6700\u5c0f\u5316\u76ee\u6807\u51fd\u6570\u7684\u4f18\u5316\u65b9\u6cd5\u7684\u4e0d\u540c\uff0c\u4f1a\u6709\u4e0d\u540c\u7684\u7ebf\u6027\u56de\u5f52\u7b97\u6cd5\u3002<\/p>\n

  \u7531\u4e8e\u5176\u4ed6\u7248\u672c\u7684\u7ebf\u6027\u56de\u5f52\u6a21\u578b\u7684\u53c2\u6570\u7c7b\u4f3c\u4e8eLinearRegression<\/code>\uff0c\u5373\u5176\u4ed6\u7c7b\u578b\u7684\u7ebf\u6027\u56de\u5f52\u6a21\u578b\u7684\u53c2\u6570\u8be6\u89e3\u90fd\u4f1a\u8df3\u8fc7\uff0c\u53ea\u4f1a\u8bb2\u89e3\u5b83\u4e0eLinearRegression<\/code>\u7684\u4e0d\u540c\u4e4b\u5904\u3002\u6211\u4eec\u63a5\u4e0b\u6765\u7684\u76ee\u7684\u5c31\u662f\u4e3a\u4e86\u7ed9\u5927\u5bb6\u4ecb\u7ecdscikit-learn\u5e93\u4e2d\u5e38\u7528\u7684\u7ebf\u6027\u56de\u5f52\u6a21\u578b\u3002<\/p>\n

LinearRegression<\/h1>\n

\u4f7f\u7528\u573a\u666f<\/h2>\n

  LinearRegression<\/code>\u56de\u5f52\u6a21\u578b\uff0c\u5373\u6211\u4eec\u5728\u7ebf\u6027\u56de\u5f52\u4e2d\u8bb2\u5230\u7684\u666e\u901a\u7ebf\u6027\u56de\u5f52\uff0c\u8be5\u666e\u901a\u7ebf\u6027\u56de\u5f52\u53ef\u4ee5\u5904\u7406\u4e00\u5143\u7ebf\u6027\u56de\u5f52\uff0c\u4e5f\u53ef\u4ee5\u5904\u7406\u591a\u5143\u7ebf\u6027\u56de\u5f52\uff0c\u4f46\u662f\u8be5\u7c7b\u4f7f\u7528\u7684\u4f18\u5316\u65b9\u6cd5\u662f\u6700\u5c0f\u4e8c\u4e58\u6cd5<\/strong>\u3002<\/p>\n

  \u901a\u5e38\u60c5\u51b5\u4e0b\u8be5\u7c7b\u662f\u6211\u4eec\u4f7f\u7528\u7ebf\u6027\u56de\u5f52\u5904\u7406\u7ebf\u6027\u95ee\u9898\u7684\u9996\u9009\u65b9\u6cd5\uff0c\u56e0\u4e3a\u5b83\u7684\u76ee\u6807\u51fd\u6570\u8f83\u5176\u4ed6\u7ebf\u6027\u56de\u5f52\u7b80\u5355\uff0c\u8ba1\u7b97\u91cf\u5c0f\uff0c\u5982\u679c\u5b83\u62df\u5408\u6570\u636e\u51fa\u73b0\u8fc7\u62df\u5408\u95ee\u9898\u5219\u53ef\u4ee5\u8003\u8651\u4f7f\u7528\u6b63\u5219\u5316\u5f62\u5f0f\u7684\u7ebf\u6027\u56de\u5f52\u3002<\/p>\n

\u4ee3\u7801<\/h2>\n
import numpy as np\nfrom sklearn.linear_model import LinearRegression\n\nX = np.array([[2, 0], [1, 9], [6, 6], [8, 8]])\n# y = 1 * x_0 + 2 * x_1 + 3\ny = np.dot(X, np.array([6, 8])) + 3<\/code><\/pre>\n
reg = LinearRegression()\nreg.fit(X, y)<\/code><\/pre>\n
LinearRegression(copy_X=True, fit_intercept=True, n_jobs=None,\n         normalize=False)<\/code><\/pre>\n
reg.score(X, y)<\/code><\/pre>\n
1.0<\/code><\/pre>\n
reg.coef_<\/code><\/pre>\n
array([6., 8.])<\/code><\/pre>\n
reg.intercept_<\/code><\/pre>\n
2.999999999999986<\/code><\/pre>\n
reg.predict(np.array([[8, 6]]))<\/code><\/pre>\n
array([99.])<\/code><\/pre>\n
\u53c2\u6570\u8be6\u89e3<\/h2>\n
    \n
  • fit_intercept\uff1a<\/strong>\u622a\u8ddd(\u504f\u7f6e\u5355\u5143)\uff0cbool\u7c7b\u578b\u3002\u662f\u5426\u5b58\u5728\u622a\u8ddd\u6216\u8005\u504f\u7f6e\u5355\u5143\u3002\u5982\u679c\u4f7f\u7528\u4e2d\u5fc3\u5316\u7684\u6570\u636e(\u4e2d\u5fc3\u70b9\u4e3a0\u7684\u6570\u636e)\uff0c\u53ef\u4ee5\u8003\u8651\u8bbe\u7f6efit_intercept=False\u3002\u9ed8\u8ba4\u4e3aTrue\u3002<\/li>\n
  • normalize\uff1a<\/strong>\u6807\u51c6\u5316\u6570\u636e\uff0cbool\u7c7b\u578b\u3002\u5f53fit_intercept=False\u7684\u65f6\u5019\uff0c\u8fd9\u4e2a\u53c2\u6570\u4f1a\u88ab\u81ea\u52a8\u5ffd\u7565\uff1b\u5982\u679cfit_intercept=True\uff0c\u56de\u5f52\u5668\u4f1a\u6807\u51c6\u5316\u8f93\u5165\u6570\u636e\uff0c\u8be5\u6807\u51c6\u5316\u65b9\u5f0f\u4e3a\uff1a\u51cf\u53bb\u5e73\u5747\u503c\uff0c\u5e76\u4e14\u9664\u4ee5\u76f8\u5e94\u7684\u4e8c\u8303\u6570\u3002\u5efa\u8bae\u5728\u4f7f\u7528fit()\u8bad\u7ec3\u6a21\u578b\u4e4b\u524d\u4f7f\u7528sklearn.preprocessing.StandardScaler\u5bf9\u6570\u636e\u6807\u51c6\u5316\uff0c\u540c\u65f6\u8bbe\u7f6enormalize=False\u3002\u9ed8\u8ba4\u4e3aFalse\u3002<\/li>\n
  • copy_X\uff1a<\/strong>\u590d\u5236\u6570\u636e\uff0cbool\u7c7b\u578b\u3002\u5982\u679ccopy_X=False\uff0c\u53ef\u80fd\u4f1a\u56e0\u4e3a\u5bf9\u6570\u636e\u4e2d\u5fc3\u5316\u628a\u539f\u59cbX\u6570\u636e\u8986\u76d6\u3002\u9ed8\u8ba4\u4e3aTrue\u3002<\/li>\n
  • n_jobs\uff1a<\/strong>\u5e76\u884c\u6570\uff0cint\u7c7b\u578b\u3002n_jobs=1\u4f7f\u75281\u4e2acpu\u8fd0\u884c\u7a0b\u5e8f\uff1bn_jobs=2\uff0c\u4f7f\u75282\u4e2acpu\u8fd0\u884c\u7a0b\u5e8f\uff1bn_jobs=-1\uff0c\u4f7f\u7528\u6240\u6709cpu\u8fd0\u884c\u7a0b\u5e8f\u3002\u9ed8\u8ba4\u4e3a1\u3002<\/li>\n<\/ul>\n
    \u5c5e\u6027<\/h2>\n
      \n
    • coef_\uff1a<\/strong>array\u7c7b\u578b\uff0c\u7ebf\u6027\u56de\u5f52\u7cfb\u6570\u3002<\/li>\n
    • intercept_\uff1a<\/strong>array\u7c7b\u578b\uff0c\u622a\u8ddd\u3002<\/li>\n<\/ul>\n
      \u65b9\u6cd5<\/h2>\n
        \n
      • fit(X,y,sample_weight=None)\uff1a<\/strong>\u628a\u6570\u636e\u653e\u5165\u6a21\u578b\u4e2d\u8bad\u7ec3\u6a21\u578b\uff0c\u5176\u4e2dsample_weight=None<\/strong>\u662farray\u7c7b\u578b\u53ef\u4ee5\u5bf9\u8bad\u7ec3\u96c6\u4e2d\u5b9e\u4f8b\u6dfb\u52a0\u6743\u91cd\uff0c\u5373\u5bf9\u8bad\u7ec3\u96c6\u4e2d\u4e0d\u540c\u7684\u6570\u636e\u589e\u52a0\u4e0d\u540c\u7684\u6743\u91cd\u3002<\/li>\n
      • get_params([deep])\uff1a<\/strong>\u8fd4\u56de\u6a21\u578b\u7684\u53c2\u6570\uff0c\u4f8b\u5982\u53ef\u4ee5\u7528\u4e8ePipeline\u4e2d\u3002<\/li>\n<\/ul>\n
        from sklearn.pipeline import Pipeline  \n\np =Pipeline([  \n        ('poly', PolynomialFeatures()),  \n        ('linear', LinearRegression(fit_intercept=False))])\nlin = p.get_params('linear')['linear']  \nprint(lin.coef_)<\/code><\/pre>\n
          \n
        • predict(X)\uff1a<\/strong>\u901a\u8fc7\u6837\u672cX\u5f97\u5230X\u5bf9\u5e94\u7684\u9884\u6d4b\u503c\u3002<\/li>\n
        • score(X, y[, sample_weight])\uff1a<\/strong>\u57fa\u4e8e\u62a5\u544a\u51b3\u5b9a\u7cfb\u6570$R^2$\u8bc4\u4f30\u6a21\u578b\u3002<\/li>\n
        • set_prams(**params)\uff1a<\/strong>\u521b\u5efa\u6a21\u578b\u53c2\u6570\u3002<\/li>\n<\/ul>\n
          \u62a5\u544a\u51b3\u5b9a\u7cfb\u6570<\/h3>\n
            \u62a5\u544a\u51b3\u5b9a\u7cfb\u6570$(R^2)$\uff0c\u53ef\u4ee5\u7406\u89e3\u6210MSE\u7684\u6807\u51c6\u7248\uff0c$R^2$\u7684\u516c\u5f0f\u4e3a
          \n$$
          \nR^2 = 1-{\\frac {{\\frac{1}{n}\\sum{i=1}^n(y^{(i)}-\\hat{y^{(i)}})^2}} {{\\frac{1}{n}}\\sum<\/em>{i=1}^n(y^{(i)}-\\mu{(y)})^2} }
          \n$$
          \n\u5176\u4e2d$\\mu<\/em>{(y)}$\u662f$y$\u7684\u5e73\u5747\u503c\uff0c\u5373${{\\frac{1}{n}}\\sum{i=1}^n(y^{(i)}-\\mu<\/em>{(y)})^2}$\u4e3a$y$\u7684\u65b9\u5dee\uff0c\u516c\u5f0f\u53ef\u4ee5\u5199\u6210
          \n$$
          \nR^2 = 1-{\\frac{MSE}{Var(y)}}
          \n$$
          \n  $R^2$\u7684\u53d6\u503c\u8303\u56f4\u5728$0-1$\u4e4b\u95f4\uff0c\u5982\u679c$R^2=1$\uff0c\u5219\u5747\u65b9\u8bef\u5dee$MSE=0$\uff0c\u5373\u6a21\u578b\u5b8c\u7f8e\u7684\u62df\u5408\u6570\u636e\u3002
          \n
          \"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"<\/div><\/p>\n
          ARDRegression<\/h1>\n
            \u5f53\u6570\u636e\u96c6\u4e2d\u6709\u5f88\u591a\u7f3a\u5931\u503c\u6216\u5f02\u5e38\u503c\u65f6\u4f7f\u7528ARDRegression<\/code>\u6a21\u578b\uff0c\u8be5\u6a21\u578b\u5c5e\u4e8e\u8d1d\u53f6\u65af\u56de\u5f52\u6a21\u578b\u3002\u8be5\u6a21\u578b\u4f1a\u5bf9\u6a21\u578b\u8f93\u51fa$Y$\u548c\u6a21\u578b\u53c2\u6570$\\omega$\u4f5c\u51fa\u5206\u5e03\u5047\u8bbe\uff0c\u5e76\u4e14\u6b63\u5219\u5316\u53c2\u6570alpha<\/strong>\u4e5f\u4f1a\u4ece\u6570\u636e\u4e2d\u4f30\u8ba1\u5f97\u5230\uff0c\u867d\u7136\u8be5\u6a21\u578b\u5bf9\u5f02\u5e38\u503c\u9c81\u68d2\u6027\u5f88\u597d\uff0c\u4f46\u7531\u4e8e\u8be5\u6a21\u578b\u8ba1\u7b97\u91cf\u5927\uff0c\u8017\u65f6\uff0c\u4e00\u822c\u60c5\u51b5\u4e0d\u63a8\u8350\u4f7f\u7528\uff0c\u6b64\u5904\u4e0d\u591a\u8d58\u8ff0\u3002<\/p>\n
          BayesianRidge<\/h1>\n
            \u8be5\u6a21\u578b\u7c7b\u4f3c\u4e8eARDRegression<\/code>\u6a21\u578b\uff0c\u4e24\u8005\u90fd\u5c5e\u4e8e\u8d1d\u53f6\u65af\u56de\u5f52\uff0c\u4e0d\u540c\u4e4b\u5904\u5728\u4e8e\u5bf9$\\omega$\u7684\u5206\u5e03\u5047\u8bbe\u4e0d\u540c\u3002\u7531\u4e8e\u8be5\u6a21\u578b\u7684\u76ee\u6807\u51fd\u6570\u7c7b\u4f3c\u4e8eRidge<\/code>\u6a21\u578b\u7684\u76ee\u6807\u51fd\u6570\uff0c\u56e0\u6b64\u53d6\u540dBayesianRidge<\/code>\u3002\u4f46\u7531\u4e8e\u8be5\u6a21\u578b\u540c\u6837\u8ba1\u7b97\u91cf\u5927\uff0c\u8017\u65f6\uff0c\u4e00\u822c\u60c5\u51b5\u4e0b\u4e5f\u4e0d\u63a8\u8350\u4f7f\u7528\uff0c\u6b64\u5904\u4e0d\u591a\u8d58\u8ff0\u3002<\/p>\n
          ElasticNet<\/h1>\n
            ElasticNet<\/code>\u6a21\u578b\u7684\u4f18\u5316\u65b9\u6cd5\u662f\u5750\u6807\u8f74\u4e0b\u964d\u6cd5<\/strong>\uff0c\u8be5\u6a21\u578b\u7531L1\u6b63\u5219\u5316\u548cL2\u6b63\u5219\u5316\u7684\u52a0\u6743\u5f97\u5230\uff0c\u5982\u679c\u4f7f\u7528L1\u6b63\u5219\u5316\u548cL2\u6b63\u5219\u5316\u90fd\u4e0d\u884c\u7684\u65f6\u5019\uff0c\u53ef\u4ee5\u8003\u8651\u4f7f\u7528\u8be5\u6a21\u578b\u3002<\/p>\n
            \u8be5\u6a21\u578b\u7531\u4e8e\u589e\u52a0\u4e86\u53c2\u6570alpha<\/strong>\u548cl1_ratio<\/strong>\uff0c\u9700\u8981\u624b\u52a8\u8c03\u53c2\uff0c\u901a\u5e38\u4f7f\u7528\u63a5\u4e0b\u6765\u7684LassoCV<\/code>\u3002<\/p>\n
          ElasticNetCV<\/h1>\n
            ElasticNetCV<\/code>\u6a21\u578b\u5728\u76ee\u6807\u51fd\u6570\u548c\u4f18\u5316\u65b9\u5f0f\u7c7b\u4f3c\u4e8eElasticNet<\/code>\uff0c\u4f46\u662f\u53ef\u4ee5\u81ea\u5df1\u624b\u52a8\u8f93\u516510\u7ec4\u3001100\u7ec4\u53c2\u6570alpha<\/strong>\u548cl1_ratio<\/strong>\uff0c\u8be5\u6a21\u578b\u4f1a\u901a\u8fc7\u4ea4\u53c9\u9a8c\u8bc1\u540e\u7ed9\u4f60\u8fd9\u7ec4\u53c2\u6570\u4e2d\u6700\u4f18\u6a21\u578b\u3002<\/p>\n
          Lasso<\/h1>\n
            Lasso<\/code>\u6a21\u578b\u7684\u4f18\u5316\u65b9\u6cd5\u662f\u5750\u6807\u8f74\u4e0b\u964d\u6cd5<\/strong>\uff0c\u8be5\u6a21\u578b\u5373\u7ebf\u6027\u56de\u5f52L1\u6b63\u5219\u5316\uff0c\u8be5\u3002\u5982\u679c\u6570\u636e\u96c6\u7684\u7279\u5f81\u7ef4\u5ea6\u8f83\u9ad8\uff0c\u53ef\u4ee5\u4f7f\u7528\u8be5\u6a21\u578b\uff0c\u8be5\u6a21\u578b\u53ef\u4ee5\u628a\u4e00\u4e9b\u8f83\u5c0f\u7684\u56de\u5f52\u7cfb\u6570\u76f4\u63a5\u53d8\u4e3a$0$\uff0c\u7531\u4e8e\u51cf\u5c11\u4e86\u6570\u636e\u96c6\u7684\u7279\u5f81\u7ef4\u5ea6\uff0c\u4e5f\u4f1a\u95f4\u63a5\u7684\u51cf\u8f7b\u6a21\u578b\u8fc7\u62df\u5408\u95ee\u9898\uff0c\u589e\u5f3a\u6a21\u578b\u7684\u6cdb\u5316\u80fd\u529b\u3002<\/p>\n
            \u8be5\u6a21\u578b\u7531\u4e8e\u4f1a\u628a\u4e00\u4e9b\u8f83\u5c0f\u7684\u56de\u5f52\u7cfb\u6570\u53d8\u4e3a$0$\uff0c\u65e2\u53ef\u4ee5\u627e\u51fa\u91cd\u8981\u7684\u7279\u5f81\uff0c\u5bf9\u6570\u636e\u96c6\u7684\u89e3\u91ca\u80fd\u529b\u5f3a\u3002<\/p>\n
            \u8be5\u6a21\u578b\u7531\u4e8e\u589e\u52a0\u4e86\u53c2\u6570alpha<\/strong>\uff0c\u9700\u8981\u624b\u52a8\u8c03\u53c2\uff0c\u901a\u5e38\u4f7f\u7528\u63a5\u4e0b\u6765\u7684LassoCV<\/code>\u3002<\/p>\n
          LassoCV<\/h1>\n
            LassoCV<\/code>\u6a21\u578b\u5728\u76ee\u6807\u51fd\u6570\u548c\u4f18\u5316\u65b9\u5f0f\u7c7b\u4f3c\u4e8eLasso<\/code>\uff0c\u4f46\u662f\u53ef\u4ee5\u81ea\u5df1\u624b\u52a8\u8f93\u516510\u7ec4\u3001100\u7ec4\u53c2\u6570alpha<\/strong>\uff0c\u8be5\u6a21\u578b\u4f1a\u901a\u8fc7\u4ea4\u53c9\u9a8c\u8bc1\u540e\u7ed9\u4f60\u8fd9\u7ec4\u53c2\u6570\u4e2d\u6700\u4f18\u6a21\u578b\u3002<\/p>\n
          LassoLars<\/h1>\n
            LassoLars<\/code>\u6a21\u578b\u7684\u4f18\u5316\u65b9\u6cd5\u662f\u6700\u5c0f\u89d2\u56de\u5f52\u6cd5<\/strong>\uff0c\u8be5\u6a21\u578b\u7c7b\u4f3c\u4e8eLasso<\/code>\u6a21\u578b\uff0c\u4f46\u662f\u8be5\u6a21\u578b\u4f18\u5316\u65b9\u6cd5\u4e3a\u3002<\/p>\n
            \u8be5\u6a21\u578b\u7531\u4e8e\u589e\u52a0\u4e86\u53c2\u6570alpha<\/strong>\uff0c\u9700\u8981\u624b\u52a8\u8c03\u53c2\uff0c\u901a\u5e38\u4f7f\u7528\u63a5\u4e0b\u6765\u7684LassoLarsCV<\/code>\u3002<\/p>\n
          LassoLarsCV<\/h1>\n
            LassoLarsCV<\/code>\u6a21\u578b\u5728\u76ee\u6807\u51fd\u6570\u548c\u4f18\u5316\u65b9\u5f0f\u7c7b\u4f3c\u4e8eLassoLars<\/code>\uff0c\u4f46\u662f\u53ef\u4ee5\u81ea\u5df1\u624b\u52a8\u8f93\u516510\u7ec4\u3001100\u7ec4\u53c2\u6570alpha<\/strong>\uff0c\u8be5\u6a21\u578b\u4f1a\u901a\u8fc7\u4ea4\u53c9\u9a8c\u8bc1\u540e\u7ed9\u4f60\u8fd9\u7ec4\u53c2\u6570\u4e2d\u6700\u4f18\u6a21\u578b\u3002<\/p>\n
          LassoLarsIC<\/h1>\n
            LassoLarsIC<\/code>\u6a21\u578b\u7c7b\u4f3c\u4e8eLasso<\/code>\u6a21\u578b\uff0c\u4e0d\u540c\u4e4b\u5904\u5728\u4e8e\u5b83\u5e76\u4e0d\u4f7f\u7528\u4ea4\u53c9\u9a8c\u8bc1\u7684\u65b9\u5f0f\u5f97\u5230\u6700\u4f18\u6a21\u578b\u3002\u5b83\u57fa\u4e8eAIC\u548cBIC\u51c6\u5219\uff0c\u4e00\u8f6e\u5c31\u53ef\u4ee5\u627e\u5230\u627e\u5230\u4e00\u4e2a\u6700\u4f18alpha<\/strong>\u548c\u6700\u4f18\u6a21\u578b\uff0c\u800c\u4ea4\u53c9\u9a8c\u8bc1\u5982\u679c\u4f7f\u7528$k$\u6298\u4ea4\u53c9\u9a8c\u8bc1\uff0c\u5219\u9700\u8981$k-1$\u6b21\u624d\u80fd\u627e\u5230\u6700\u4f18\u6a21\u578b\u3002<\/p>\n
            \u8be5\u6a21\u578b\u4ece\u4e0a\u8ff0\u8bb2\u8ff0\u770b\u8d77\u6765\u662f\u5f88\u5b8c\u7f8e\u7684\uff0c\u4f46\u662f\u8be5\u6a21\u578b\u8981\u6c42\u6570\u636e\u96c6\u662f\u7531\u67d0\u4e2a\u5047\u8bbe\u7684\u6a21\u578b\u4ea7\u751f\u7684\uff0c\u5e76\u4e14\u5982\u679c\u5f53\u7279\u5f81\u6570\u91cf\u5927\u4e8e\u5b9e\u4f8b\u6570\u91cf\u7684\u65f6\u5019\u8be5\u6a21\u578b\u53ef\u80fd\u4f1a\u6210\u4e3a\u4e00\u4e2a\u8f83\u5dee\u7684\u6a21\u578b\uff0c\u6240\u4ee5\u5728\u5de5\u4e1a\u4e0a\u4e00\u822c\u4e0d\u63a8\u8350\u4f7f\u7528\u3002<\/p>\n
          MutilTaskLasso<\/h1>\n
            MutilTaskLasso<\/code>\u6a21\u578b\u7684\u4f18\u5316\u65b9\u6cd5\u662f\u5750\u6807\u8f74\u4e0b\u964d\u6cd5<\/strong>\uff0c\u6a21\u578b\u4e2d\u7684MutilTask\u53ef\u4ee5\u7406\u89e3\u6210\u201c\u591a\u4e2a\u201d\u800c\u4e0d\u662f\u201c\u591a\u8fdb\u7a0b\u201d\uff0c\u5373\u4e00\u6b21\u6027\u4f7f\u7528\u591a\u4e2aL1\u6b63\u5219\u5316\u7ebf\u6027\u56de\u5f52\u6a21\u578b\u62df\u5408\u6570\u636e\uff0c\u6709\u65f6\u5019\u4e5f\u79f0\u4e4b\u4e3a\u5171\u4eab\u7279\u5f81\u534f\u540c\u56de\u5f52\u3002<\/p>\n
            \u666e\u901a\u7ebf\u6027\u56de\u5f52\u7684\u6a21\u578b\u662f
          \n$$
          \nY = X\\omega
          \n$$
          \n\u5176\u4e2d\u5047\u8bbe\u4e00\u4e2a\u6570\u636e\u96c6\u6709$m$\u5b9e\u4f8b\uff0c\u6bcf\u4e2a\u5b9e\u4f8b\u6709$n$\u4e2a\u7279\u5f81\uff0c\u5219\u5176\u4e2d$Y$\u7684\u7ef4\u5ea6\u662f$m1$\uff0c$X$\u7684\u7ef4\u5ea6\u662f$m<\/em>n$\uff0c$\\omega$\u7684\u7ef4\u5ea6\u662f$n*1$\u3002<\/p>\n
            \u8be5\u6a21\u578b\u53bb\u6389\u6b63\u5219\u5316\u9879\u662f
          \n$$
          \nY = XW
          \n$$
          \n\u5176\u4e2d\u5047\u8bbe\u4e00\u4e2a\u6570\u636e\u96c6\u6709$m$\u5b9e\u4f8b\uff0c\u6bcf\u4e2a\u5b9e\u4f8b\u6709$n$\u4e2a\u7279\u5f81\uff0c\u5219\u5176\u4e2d$Y$\u7684\u7ef4\u5ea6\u662f$mk$\uff0c$X$\u7684\u7ef4\u5ea6\u662f$m<\/em>n$\uff0c$W$\u7684\u7ef4\u5ea6\u662f$n*k$\uff0c\u5176\u4e2d$k$\u4e3a\u56de\u5f52\u6a21\u578b\u7684\u4e2a\u6570\uff0c\u5373\u8be5\u6a21\u578b\u7684fit()\u65b9\u6cd5\u53ef\u4ee5\u4f20\u5165$k$\u7ef4\u7684\u7279\u5f81\u3002<\/p>\n
            \u8be5\u6a21\u578b\u7531\u4e8e\u589e\u52a0\u4e86\u53c2\u6570alpha<\/strong>\u548c$k$\uff0c\u9700\u8981\u624b\u52a8\u8c03\u53c2\uff0c\u901a\u5e38\u4f7f\u7528\u63a5\u4e0b\u6765\u7684MutilTaskLassoCV<\/code>\u3002<\/p>\n
          MutilTaskElasticNet<\/h1>\n
            MutilTaskElasticNet<\/code>\u6a21\u578b\u7684\u4f18\u5316\u65b9\u6cd5\u662f\u5750\u6807\u8f74\u4e0b\u964d\u6cd5<\/strong>\uff0c\u8be5\u6a21\u578b\u7c7b\u4f3c\u4e8eMutilTaskLasso<\/code>\u6a21\u578b\uff0c\u53ea\u662f\u5728\u6b63\u5219\u5316\u9879\u4e0a\u524d\u8005\u4f7f\u7528\u4e86L1\u6b63\u5219\u9879\uff0c\u540e\u8005\u4f7f\u7528\u4e86\u5f39\u6027\u7f51\u7edc\u6b63\u5219\u9879\u3002<\/p>\n
            \u8be5\u6a21\u578b\u7531\u4e8e\u589e\u52a0\u4e86\u53c2\u6570alpha<\/strong>\u548cl1_ratio<\/strong>\uff0c\u9700\u8981\u624b\u52a8\u8c03\u53c2\uff0c\u901a\u5e38\u4f7f\u7528\u63a5\u4e0b\u6765\u7684LassoCV<\/code>\u3002<\/p>\n
          MutilTaskLassoCV<\/h1>\n
            MutilTaskLassoCV<\/code>\u6a21\u578b\u5728\u76ee\u6807\u51fd\u6570\u548c\u4f18\u5316\u65b9\u5f0f\u7c7b\u4f3c\u4e8eMutilTaskLasso<\/code>\uff0c\u4f46\u662f\u53ef\u4ee5\u81ea\u5df1\u624b\u52a8\u8f93\u516510\u7ec4\u3001100\u7ec4\u53c2\u6570alpha<\/strong>\uff0c\u8be5\u6a21\u578b\u4f1a\u901a\u8fc7\u4ea4\u53c9\u9a8c\u8bc1\u540e\u7ed9\u4f60\u8fd9\u7ec4\u53c2\u6570\u4e2d\u6700\u4f18\u6a21\u578b\u3002<\/p>\n
          MutilTaskElasticNetCV<\/h1>\n
            \u8be5\u6a21\u578b\u5728\u76ee\u6807\u51fd\u6570\u548c\u4f18\u5316\u65b9\u5f0f\u7c7b\u4f3c\u4e8eLasso<\/code>\uff0c\u4f46\u662f\u53ef\u4ee5\u81ea\u5df1\u624b\u52a8\u8f93\u516510\u7ec4\u3001100\u7ec4\u53c2\u6570alpha<\/strong>\u548cl1_ratio<\/strong>\uff0c\u8be5\u6a21\u578b\u4f1a\u901a\u8fc7\u4ea4\u53c9\u9a8c\u8bc1\u540e\u7ed9\u4f60\u8fd9\u7ec4\u53c2\u6570\u4e2d\u6700\u4f18\u6a21\u578b\u3002<\/p>\n
          OrthogonalMatchingPursuit<\/h1>\n
            OrthogonalMatchingPursuit<\/code>\u6a21\u578b\u4f18\u5316\u65b9\u6cd5\u662f\u524d\u5411\u9009\u62e9\u7b97\u6cd5<\/strong>\uff0c\u4f18\u5316\u65b9\u6cd5\u901f\u5ea6\u867d\u7136\u5feb\uff0c\u4f46\u662f\u7cbe\u786e\u5ea6\u8f83\u4f4e\u3002<\/p>\n
            \u8be5\u6a21\u578b\u4f7f\u7528\u53c2\u6570n_nonzero_coefs<\/strong>\u9650\u5236\u6a21\u578b\u53c2\u6570$\\omega$\u5411\u91cf\u4e2d\u5143\u7d20\u975e$0$\u7684\u4e2a\u6570\uff0c\u7531\u4e8e\u8be5\u7279\u5f81\u53ef\u4ee5\u7528\u4e8e\u7a00\u758f\u7279\u5f81\u6a21\u578b\u7684\u7279\u5f81\u9009\u62e9\u4e0a\uff0c\u8fd9\u4e00\u70b9\u7c7b\u4f3c\u4e8eLasso<\/code>\u6a21\u578b\uff0c\u4f46\u662f\u7531\u4e8e\u4f18\u5316\u65b9\u6cd5\u524d\u5411\u9009\u62e9\u7b97\u6cd5<\/strong>\uff0c\u4e00\u822c\u4e0d\u63a8\u8350\u4f7f\u7528\u3002<\/p>\n
            \u8be5\u6a21\u578b\u7531\u4e8e\u589e\u52a0\u4e86\u53c2\u6570n_nonzero_coefs<\/strong>\uff0c\u9700\u8981\u624b\u52a8\u8c03\u53c2\uff0c\u901a\u5e38\u4f7f\u7528\u63a5\u4e0b\u6765\u7684LassoCV<\/code>\u3002<\/p>\n
          OrthogonalMatchingPursuitCV<\/h1>\n
            OrthogonalMatchingPursuitCV<\/code>\u6a21\u578b\u5728\u76ee\u6807\u51fd\u6570\u548c\u4f18\u5316\u65b9\u5f0f\u7c7b\u4f3c\u4e8eOrthogonalMatchingPursuitCV<\/code>\uff0c\u4f46\u662f\u53ef\u4ee5\u81ea\u5df1\u624b\u52a8\u8f93\u516510\u7ec4\u3001100\u7ec4\u53c2\u6570n_nonzero_coefs<\/strong>\uff0c\u8be5\u6a21\u578b\u4f1a\u901a\u8fc7\u4ea4\u53c9\u9a8c\u8bc1\u540e\u7ed9\u4f60\u8fd9\u7ec4\u53c2\u6570\u4e2d\u6700\u4f18\u6a21\u578b\u3002<\/p>\n
          RANSACRegressor<\/h1>\n
             RANSACRegressor<\/code>\u6a21\u578b\u4f7f\u7528\u7684\u4f18\u5316\u7b97\u6cd5\u662fRANSACR\u7b97\u6cd5\uff0c\u8be5\u7b97\u6cd5\u53ef\u4ee5\u63a7\u5236\u4f7f\u7528\u90e8\u5206\u533a\u57df\u7684\u6570\u636e\u96c6\u8bad\u7ec3\u6a21\u578b\u3002<\/p>\n
            \u53ef\u4ee5\u53c2\u8003\u300aRANSAC\u7b97\u6cd5\u7ebf\u6027\u56de\u5f52(\u6ce2\u65af\u987f\u623f\u4ef7\u9884\u6d4b)\u300b<\/strong>\u3002<\/p>\n
          Ridge<\/h1>\n
            Ridge<\/code>\u6a21\u578b\u7684\u4f18\u5316\u65b9\u6cd5\u662f\u6700\u5c0f\u4e8c\u4e58\u6cd5<\/strong>\uff0c\u8be5\u6a21\u578b\u5373\u7ebf\u6027\u56de\u5f52L2\u6b63\u5219\u5316\uff0c\u4e00\u822c\u4f7f\u7528LinearRegression<\/code>\u6a21\u578b\u65f6\u6a21\u578b\u8fc7\u62df\u5408\u65f6\u53ef\u4ee5\u4f7f\u7528\u8be5\u65b9\u6cd5\u3002<\/p>\n
            \u7531\u4e8e\u989d\u5916\u589e\u52a0\u4e86alpha<\/strong>\u53c2\u6570\uff0c\u4e00\u822c\u60c5\u51b5\u4e0b\u9700\u8981\u81ea\u5df1\u624b\u52a8\u8c03\u53c2\uff0c\u6240\u4ee5\u53ef\u4ee5\u5728\u81ea\u5df1\u6d4b\u8bd5\u7684\u65f6\u5019\u4f7f\u7528\uff0c\u4e00\u822c\u5de5\u4e1a\u4e0a\u4f7f\u7528\u8f83\u591a\u7684\u662f\u63a5\u4e0b\u6765\u7684RidgeCV<\/code>\u3002<\/p>\n
          RidgeCV<\/h1>\n
            RidgeCV<\/code>\u6a21\u578b\u5728\u76ee\u6807\u51fd\u6570\u548c\u4f18\u5316\u65b9\u5f0f\u7c7b\u4f3c\u4e8eRidge<\/code>\uff0c\u4f46\u662f\u53ef\u4ee5\u81ea\u5df1\u624b\u52a8\u8f93\u516510\u7ec4\u3001100\u7ec4\u53c2\u6570alpha<\/strong>\uff0c\u8be5\u6a21\u578b\u4f1a\u901a\u8fc7\u4ea4\u53c9\u9a8c\u8bc1\u540e\u7ed9\u4f60\u8fd9\u7ec4\u53c2\u6570\u4e2d\u6700\u4f18\u6a21\u578b\u3002
          \n
          \"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"<\/div><\/p>\n","protected":false},"excerpt":{"rendered":"
          scikit-learn\u5e93\u4e4b\u7ebf\u6027\u56de\u5f52   \u7531\u4e8escikit-learn\u5e93\u4e2dsclearn. […]<\/p>\n","protected":false},"author":3,"featured_media":2947,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[276,286],"tags":[],"_links":{"self":[{"href":"https:\/\/egonlin.com\/index.php?rest_route=\/wp\/v2\/posts\/2979"}],"collection":[{"href":"https:\/\/egonlin.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/egonlin.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/egonlin.com\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/egonlin.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2979"}],"version-history":[{"count":0,"href":"https:\/\/egonlin.com\/index.php?rest_route=\/wp\/v2\/posts\/2979\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/egonlin.com\/index.php?rest_route=\/wp\/v2\/media\/2947"}],"wp:attachment":[{"href":"https:\/\/egonlin.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2979"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/egonlin.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2979"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/egonlin.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2979"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}