{"id":62,"date":"2018-07-06T20:47:13","date_gmt":"2018-07-06T12:47:13","guid":{"rendered":"http:\/\/www.halve.top\/?p=62"},"modified":"2018-07-10T18:36:23","modified_gmt":"2018-07-10T10:36:23","slug":"%e4%bd%bf%e7%94%a8%e6%9c%ba%e5%99%a8%e5%ad%a6%e4%b9%a0%e9%a2%84%e6%b5%8b%e4%b8%96%e7%95%8c%e6%9d%af%e8%b5%9b%e6%9e%9c","status":"publish","type":"post","link":"http:\/\/www.halve.top\/?p=62","title":{"rendered":"\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u9884\u6d4b\u4e16\u754c\u676f\u8d5b\u679c"},"content":{"rendered":"<h1>\u76ee\u6807<\/h1>\n<p>\u672c\u6587\u7684\u76ee\u6807\u662f\u5728\u5b66\u4e60\u4e86TensorFlow\u7684\u5b98\u65b9<a href="\""https:\/\/www.tensorflow.org\/get_started\/get_started_for_beginners\">Graph Execution\u793a\u4f8b<\/a>\u540e\uff0c\u901a\u8fc7\u4e3e\u4e00\u53cd\u4e09\u8fd0\u7528\u5176\u4ed6\u4f8b\u5b50\u6765\u8fdb\u4e00\u6b65\u7406\u89e3\u5b98\u65b9\u793a\u4f8b\u3002\u56e0\u6b64\uff0c\u672c\u6587\u4e2d\u7684\u6240\u6709\u7ed3\u679c\u4ec5\u4f5c\u5b66\u672f\u7814\u7a76\u7528\u9014\u3002<\/p>\n<h1>\u524d\u63d0<\/h1>\n<h3>IDE\u51c6\u5907<\/h2>\n<p>Visual Studio 2017<\/p>\n<p>\u6253\u5f00Visual Studio Installer<strong>\u4fee\u6539<\/strong>\u5f53\u524d\u4f7f\u7528\u7684VS\uff0c\u786e\u4fdd<strong>\u5de5\u4f5c\u8d1f\u8f7d<\/strong>\u4e2d<strong>Web\u548c\u4e91\u5206\u7c7b<\/strong>\u4e0b\u7684<strong>Python\u5f00\u53d1<\/strong>\u88ab\u9009\u4e2d\u3002\u7136\u540e\u5207\u6362\u5230\u5355\u4e2a\u7ec4\u4ef6\uff0c\u786e\u4fdd<strong>\u5f00\u53d1\u6d3b\u52a8<\/strong>\u4e2d\u7684<strong>Python\u8bed\u8a00\u652f\u6301<\/strong>\u548c<strong>\u7f16\u8bd1\u5668\u3001\u751f\u6210\u5de5\u5177\u548c\u8fd0\u884c\u65f6<\/strong>\u4e2d\u7684<strong>Python 3 64-bit<\/strong>\u88ab\u52fe\u9009\u3002<\/p>\n<p>\u70b9\u51fb\u53f3\u4e0b\u89d2\u7684\u4fee\u6539\uff0c\u7b49\u5f85\u7ec4\u4ef6\u88ab\u4e0b\u8f7d\u548c\u5b89\u88c5\u3002<\/p>\n<h3>\u9879\u76ee\u51c6\u5907<\/h2>\n<p>\u65b0\u5efa\u4e00\u4e2a<strong>Python\u5e94\u7528\u7a0b\u5e8f<\/strong>\u9879\u76ee\u3002\u5728\u89e3\u51b3\u65b9\u6848\u8d44\u6e90\u7ba1\u7406\u5668\u4e2d\u627e\u5230<strong>Python\u73af\u5883<\/strong>\uff0c\u5e76\u53f3\u952e\u6dfb\u52a0\u865a\u62df\u73af\u5883\u3002\u865a\u62df\u73af\u5883\u7684\u57fa\u7840\u89e3\u6790\u5668\u4e3aPython 3\u4e2d\u7684\u67d0\u4e2a\u7248\u672c\u3002<\/p>\n<p>\u53f3\u952e\u70b9\u51fb\u65b0\u5efa\u7684\u865a\u62df\u73af\u5883\uff0c\u9009\u62e9<strong>\u5b89\u88c5Python\u5305<\/strong>\u3002<\/p>\n<p>\u5206\u522b\u5b89\u88c5pip\u3001pandas\u548ctensorflow\u3002\u8fd9\u4e9b\u5305\u5728\u5b89\u88c5\u7684\u540c\u65f6\uff0c\u4f1a\u628a\u6240\u6709\u9700\u8981\u7684\u5305\u8fdb\u884c\u94fe\u5f0f\u5b89\u88c5\u3002\u4ee5\u9632\u4e07\u4e00\uff0c\u4ee5\u4e0b\u5217\u51fa\u9700\u8981\u7528\u5230\u7684\u6240\u6709\u5305\u53ca\u7248\u672c\u3002<\/p>\n<pre class=\"brush: plain; collapse: true; light: false; title: requirements.txt; toolbar: true; notranslate\" title=\"requirements.txt\">\r\nabsl-py==0.2.2\r\nastor==0.6.2\r\nbleach==1.5.0\r\ncertifi==2018.4.16\r\nchardet==3.0.4\r\ncycler==0.10.0\r\ngast==0.2.0\r\nget==1.0.3\r\ngrpcio==1.13.0\r\nhtml5lib==0.9999999\r\nidna==2.7\r\nkiwisolver==1.0.1\r\nMarkdown==2.6.11\r\nmatplotlib==2.2.2\r\nnumpy==1.14.5\r\npandas==0.23.1\r\npip==10.0.1\r\npost==1.0.2\r\nprotobuf==3.6.0\r\npublic==1.0.3\r\npyparsing==2.2.0\r\npython-dateutil==2.7.3\r\npytz==2018.5\r\nquery-string==1.0.2\r\nrequest==1.0.2\r\nrequests==2.19.1\r\nsetuptools==39.2.0\r\nsix==1.11.0\r\ntensorboard==1.8.0\r\ntensorflow==1.8.0\r\ntermcolor==1.1.0\r\nurllib3==1.23\r\nWerkzeug==0.14.1\r\nwheel==0.31.1\r\n<\/pre>\n<h3>\u6570\u636e\u51c6\u5907<\/h2>\n<p>\u65e2\u7136\u8981\u8fdb\u884c\u673a\u5668\u5b66\u4e60\uff0c\u5c31\u9700\u8981\u51c6\u5907\u597d<strong>\u8bad\u7ec3\u6570\u636e\u96c6<\/strong>\u548c<strong>\u6d4b\u8bd5\u6570\u636e\u96c6<\/strong>\u3002\u8bad\u7ec3\u6570\u636e\u96c6\u7528\u4e8e\u6784\u5efa\u51fa\u4e00\u4e2a\u80fd\u6839\u636e<strong>\u7279\u5f81\u503c<\/strong>\u63a8\u65ad\u51fa<strong>\u6807\u7b7e\u503c<\/strong>\u7684<strong>\u6a21\u578b<\/strong>\uff0c\u800c\u6d4b\u8bd5\u6570\u636e\u96c6\u7528\u4e8e<strong>\u8bc4\u4f30<\/strong>\u8fd9\u4e2a\u6a21\u578b\u7684\u51c6\u786e\u5ea6\uff0c\u5b83\u4eec\u90fd\u6765\u81ea\u4e00\u4e2a\u6570\u636e\u8d85\u96c6\u3002\u4e3a\u4e86\u6709\u6548\u5730\u5bf9\u6a21\u578b\u8fdb\u884c\u8bc4\u5224\uff0c\u6d4b\u8bd5\u6570\u636e\u96c6\u4e0e\u8bad\u7ec3\u6570\u636e\u96c6\u4e0d\u80fd\u6709\u4ea4\u96c6\u3002<\/p>\n<p>\u6211\u6784\u5efa\u7684\u6570\u636e\u8d85\u96c6\u6bcf\u4e00\u884c\u4ee3\u8868\u4e00\u573a\u6bd4\u8d5b\uff0c\u56e0\u4e3a\u6211\u5e0c\u671b\u4ee5\u4e00\u573a\u6bd4\u8d5b\u4e3a\u5355\u4f4d\u9884\u6d4b\u5b83\u7684\u8d5b\u679c\uff0c\u5e76\u6709\u5982\u4e0b\u7684\u5217\uff0c\u5e76\u63cf\u8ff0\u4e3a\u4ec0\u4e48\u6211\u8981\u7eb3\u5165\u8fd9\u4e9b\u5217\uff1a<\/p>\n<div>\n<li>\u5e8f\u53f7\uff1a\u63cf\u8ff0\u8fd9\u662f\u7b2c\u51e0\u573a\u6bd4\u8d5b\u3002\u968f\u7740\u8d5b\u7a0b\u65f6\u95f4\u7684\u8fdb\u884c\uff0c\u7403\u5458\u9010\u6e10\u9002\u5e94\u6c14\u5019\u73af\u5883\u7684\u60c5\u51b5\u4ee5\u53ca\u4ea7\u751f\u7684\u75b2\u5026\u90fd\u53ef\u80fd\u4f1a\u5f71\u54cd\u8d5b\u679c\u3002<\/li>\n<li>\u4e3b\u961f\uff1a\u4ee5\u6570\u5b57\u7684\u65b9\u5f0f\u8868\u793a\u8fd9\u573a\u6bd4\u8d5b\u7684\u4e3b\u961f\u3002\u6211\u6309\u51fa\u573a\u987a\u5e8f\u4e3a\u6bcf\u652f\u7403\u961f\u5b9a\u4e49\u4e86\u4e00\u4e2a\u7f16\u53f7\uff0c\u5b9e\u9645\u4e0a\u7f16\u53f7\u7684\u5927\u5c0f\u65e0\u6240\u8c13\uff0c\u53ea\u8981\u5728\u6574\u4e2a\u6570\u636e\u96c6\u4e2d\u7528\u540c\u4e00\u4e2a\u7f16\u53f7\u8868\u793a\u540c\u4e00\u652f\u7403\u961f\u5373\u53ef\u3002\u7f16\u53f7\u5bf9\u5e94\u7684\u7403\u961f\u5728\u540e\u9762\u4f1a\u7ed9\u51fa\u3002<\/li>\n<li>\u5ba2\u961f\uff1a\u4ee5\u6570\u5b57\u7684\u65b9\u5f0f\u8868\u793a\u8fd9\u573a\u6bd4\u8d5b\u7684\u5ba2\u961f\u3002<\/li>\n<li>\u4e3b\u961f\u80dc\u7684\u5e73\u5747\u521d\u59cb\u8d54\u7387\uff1a\u8d54\u7387\u4ee3\u8868\u7740\u67d0\u5bb6\u516c\u53f8\u5bf9\u8be5\u573a\u6bd4\u8d5b\u8be5\u961f\u8868\u73b0\u7684\u770b\u6cd5\uff0c\u5b83\u672c\u8eab\u5373\u662f\u5404\u79cd\u53ef\u80fd\u5f71\u54cd\u6bd4\u8d5b\u56e0\u7d20\u7684\u4e00\u4e2a\u7efc\u5408\u63d0\u70bc\u3002\u800c\u53d6\u591a\u5bb6\u516c\u53f8\u7684\u5e73\u5747\u8d54\u7387\u66f4\u5177\u5ba2\u89c2\u6027\u3002<\/li>\n<li>\u5e73\u5c40\u7684\u5e73\u5747\u521d\u59cb\u8d54\u7387<\/li>\n<li>\u4e3b\u961f\u8d1f\u7684\u5e73\u5c40\u521d\u59cb\u8d54\u7387<\/li>\n<li>\u4e3b\u961f\u80dc\u7684\u5e73\u5c40\u5373\u65f6\u8d54\u7387\uff1a\u901a\u8fc7\u4e0e\u540c\u4e00\u884c\u4e2d\u7684\u521d\u59cb\u8d54\u7387\u4f5c\u6bd4\u8f83\uff0c\u83b7\u5f97\u8d54\u7387\u53d8\u5316\u3002\u8fd9\u4e2a\u53d8\u5316\u53ef\u80fd\u662f\u56e0\u4e3a\u67d0\u4e9b\u4e34\u8d5b\u524d\u5f71\u54cd\u6bd4\u8d5b\u7684\u56e0\u7d20\u53d8\u5316\u6240\u5bfc\u81f4\u7684\uff0c\u4e5f\u53ef\u80fd\u662f\u516c\u53f8\u4e00\u5f00\u59cb\u5728\u63a9\u76d6\u81ea\u5df1\u7684\u771f\u5b9e\u610f\u56fe\u3002\u5373\u65f6\u8d54\u7387\u5728\u6570\u636e\u8d85\u96c6\u4e2d\u4e3a\u622a\u6b62\u6295\u6ce8\u524d\u7684\u6700\u540e\u4e00\u6b21\u516c\u5e03\u7684\u8d54\u7387\uff0c\u800c\u5728\u9884\u6d4b\u6761\u4ef6\u4e2d\u4e3a\u5f53\u524d\u5f97\u77e5\u7684\u6700\u540e\u4e00\u6b21\u516c\u5e03\u7684\u8d54\u7387\u3002<\/li>\n<li>\u5e73\u5c40\u7684\u5e73\u5747\u5373\u65f6\u8d54\u7387<\/li>\n<li>\u4e3b\u961f\u8d1f\u7684\u5e73\u5747\u5373\u65f6\u8d54\u7387<\/li>\n<li>\u4e3b\u961f\u8fd110\u573a\u6bd4\u8d5b\u7684\u80dc\u573a\u6570\uff1a\u8fd1\u671f\u6bd4\u8d5b\u4f1a\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u53cd\u6620\u51fa\u8be5\u961f\u4f0d\u7684\u72b6\u6001\u3002<\/li>\n<li>\u4e3b\u961f\u8fd110\u573a\u6bd4\u8d5b\u7684\u5e73\u573a\u6570<\/li>\n<li>\u4e3b\u961f\u8fd110\u573a\u6bd4\u8d5b\u7684\u8d1f\u573a\u6570<\/li>\n<li>\u5ba2\u961f\u8fd110\u573a\u6bd4\u8d5b\u7684\u80dc\u573a\u6570<\/li>\n<li>\u5ba2\u961f\u8fd110\u573a\u6bd4\u8d5b\u7684\u5e73\u573a\u6570<\/li>\n<li>\u5ba2\u961f\u8fd110\u573a\u6bd4\u8d5b\u7684\u8d1f\u573a\u6570<\/li>\n<li>\u4e3b\u961f\u79ef\u5206\uff1a\u8fd9\u573a\u6bd4\u8d5b\u5f00\u59cb\u524d\uff0c\u6309\u5c0f\u7ec4\u8d5b\u89c4\u5219\u7684\u79ef\u5206\u3002\u56e0\u4e3a\u6570\u636e\u96c6\u4e2d\u4e5f\u7eb3\u5165\u4e86\u6dd8\u6c70\u8d5b\u7684\u6570\u636e\uff0c\u6240\u4ee5\u4e5f\u6309\u80dc\u65b9\u79ef3\u5206\uff0c\u5e73\u5c40\u5404\u79ef1\u5206\uff0c\u8d1f\u65b9\u4e0d\u79ef\u5206\u7684\u65b9\u5f0f\u7edf\u8ba1\u3002\u79ef\u5206\u5728\u4e00\u5b9a\u7a0b\u5ea6\u4e0a\u53cd\u6620\u4e86\u4e4b\u524d\u7684\u6bd4\u8d5b\u7ed3\u679c\u5bf9\u8be5\u961f\u4f0d\u9020\u6210\u7684\u5f71\u54cd\u3002<\/li>\n<li>\u5ba2\u961f\u79ef\u5206<\/li>\n<li>\u8d5b\u679c\uff1a\u4ee5\u6570\u5b57\u65b9\u5f0f\u8868\u8ff0\u8fd9\u573a\u6bd4\u8d5b\u7684\u7ed3\u679c\u3002\u4e3b\u80dc\u4e3a2\uff0c\u5e73\u5c40\u4e3a1\uff0c\u4e3b\u8d1f\u4e3a0.<\/li>\n<\/ul>\n<p>\u524d17\u5217\u4e3a\u7279\u5f81\u5217\uff0c\u6bcf\u5217\u4e2d\u7684\u6bcf\u4e2a\u503c\u5373\u4e3a\u7279\u5f81\u503c\u3002\u6700\u540e\u4e00\u5217\u4e3a\u6807\u7b7e\u5217\uff0c\u6bcf\u5217\u4e2d\u7684\u6bcf\u4e2a\u503c\u5373\u4e3a\u6807\u7b7e\u503c\u3002\u673a\u5668\u5b66\u4e60\u7684\u76ee\u7684\u5c31\u662f\u627e\u51fa\u5df2\u77e5\u7684\u7279\u5f81\u503c\u4e0e\u7279\u5f81\u503c\u7ec4\u5408\u548c\u5df2\u77e5\u6807\u7b7e\u503c\u4e4b\u95f4\u7684\u5173\u8054\uff0c\u7136\u540e\u6839\u636e\u5df2\u77e5\u7684\u7279\u5f81\u503c\u53bb\u9884\u6d4b\u672a\u77e5\u7684\u6807\u7b7e\u3002<\/p>\n<p>\u8fd9\u4e9b\u6570\u636e\u7684\u6765\u6e90\u901a\u5e38\u53ef\u4ee5\u5728\u6295\u6ce8\u7f51\u7ad9\u627e\u5f97\u5230\u3002<\/p>\n<p>\u8fd9\u65f6\u524d\u63d0\u51c6\u5907\u5df2\u7ecf\u5b8c\u6210\uff0c\u53ef\u4ee5\u5f80\u9879\u76ee\u4e2d\u6dfb\u52a0.py\u6587\u4ef6\u53ca\u4ee3\u7801\u3002<\/p>\n<h1>\u4ee3\u7801\u5c55\u793a<\/h1>\n<p>\u5148\u4e0a\u4ee3\u7801\uff0c\u7136\u540e\u5bf9\u4ee3\u7801\u8fdb\u884c\u8bb2\u89e3\u3002<\/p>\n<pre class=\"brush: python; collapse: true; light: false; title: WorldCupPredict.py; toolbar: true; notranslate\" title=\"WorldCupPredict.py\">\r\nfrom __future__ import absolute_import\r\nfrom __future__ import division\r\nfrom __future__ import print_function\r\n\r\nimport argparse\r\nimport tensorflow as tf\r\n\r\nimport game_data\r\n\r\n\r\nparser = argparse.ArgumentParser()\r\nparser.add_argument('--batch_size', default=4, type=int, help='\u6346\u7ed1\u5927\u5c0f')\r\nparser.add_argument('--train_steps', default=2000, type=int,\r\n                    help='\u8bad\u7ec3\u6b65\u6570')\r\n\r\ndef main(argv):\r\n    args = parser.parse_args(argv[1:])\r\n\r\n    # \u62c9\u53d6\u6570\u636e\r\n    (train_x, train_y), (test_x, test_y) = game_data.load_data()\r\n\r\n    # \u63cf\u8ff0\u8f93\u5165\u7279\u5f81\u5217\r\n    my_feature_columns = []\r\n    for key in train_x.keys():\r\n        my_feature_columns.append(tf.feature_column.numeric_column(key=key))\r\n\r\n    # \u6784\u5efaDNN\u5206\u7c7b\u5668\r\n    classifier = tf.estimator.DNNClassifier(\r\n        feature_columns=my_feature_columns,\r\n        # 3\u4e2a\u9690\u85cf\u5c42\r\n        hidden_units=[20, 15, 5],\r\n        # \u8bbe\u7f6e3\u4e2a\u7c7b\u522b\r\n        n_classes=3)\r\n\r\n    # \u8bad\u7ec3\u6a21\u578b\r\n    classifier.train(\r\n        input_fn=lambda:game_data.train_input_fn(train_x, train_y,\r\n                                                 args.batch_size),\r\n        steps=args.train_steps)\r\n\r\n    # \u8bc4\u4f30\u6a21\u578b\r\n    eval_result = classifier.evaluate(\r\n        input_fn=lambda:game_data.eval_input_fn(test_x, test_y,\r\n                                                args.batch_size))\r\n\r\n    print('\\n\u6d4b\u8bd5\u96c6\u51c6\u786e\u5ea6: {accuracy:0.3f}\\n'.format(**eval_result))\r\n\r\n    # \u5efa\u7acb\u9700\u8981\u4f7f\u7528\u6a21\u578b\u505a\u51fa\u9884\u6d4b\u7684\u6570\u636e\r\n    predict_x = {\r\n        'Game': [57],\r\n        'Home': [4],\r\n        'Away': [9],\r\n        'WinInitialOdds': [4.3],\r\n        'DrawInitialOdds': [3.17],\r\n        'LossInitialOdds': [2.00],\r\n        'WinCurrentOdds': [4.44],\r\n        'DrawCurrentOdds': [3.11],\r\n        'LossCurrentOdds': [2.03],\r\n        'HomeWinCount': [8],\r\n        'HomeDrawCount': [1],\r\n        'HomeLossCount': [1],\r\n        'AwayWinCount': [6],\r\n        'AwayDrawCount': [3],\r\n        'AwayLossCount': [1],\r\n        'HomePoint': [12],\r\n        'AwayPoint': [10]\r\n    }\r\n\r\n    predictions = classifier.predict(\r\n        input_fn=lambda:game_data.eval_input_fn(predict_x,\r\n                                                labels=None,\r\n                                                batch_size=args.batch_size))\r\n\r\n    template = ('\\\u9884\u6d4b\u7ed3\u679c\u662f&quot;{}&quot; ({:.1f}%)&quot;')\r\n\r\n    for pred_dict in predictions:\r\n        class_id = pred_dict['class_ids'][0]\r\n        probability = pred_dict['probabilities'][class_id]\r\n\r\n        print(template.format(game_data.RESULT[class_id],\r\n                              100 * probability))\r\n\r\n\r\nif __name__ == '__main__':\r\n    tf.logging.set_verbosity(tf.logging.INFO)\r\n    tf.app.run(main)\r\n<\/pre>\n<pre class=\"brush: python; collapse: true; light: false; title: game_data.py; toolbar: true; notranslate\" title=\"game_data.py\">\r\nimport pandas as pd\r\nimport tensorflow as tf\r\n\r\nTRAIN_URL = &quot;http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/game_training.csv&quot;\r\nTEST_URL = &quot;http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/game_test.csv&quot;\r\n\r\nCSV_COLUMN_NAMES = ['Game','Home','Away',\r\n                    'WinInitialOdds','DrawInitialOdds','LossInitialOdds',\r\n                    'WinCurrentOdds','DrawCurrentOdds','LossCurrentOdds',\r\n                    'HomeWinCount','HomeDrawCount','HomeLossCount',\r\n                    'AwayWinCount','AwayDrawCount','AwayLossCount',\r\n                    'HomePoint','AwayPoint','Result']\r\n\r\nRESULT = ['Loss','Draw','Win']\r\n\r\ndef maybe_download():\r\n    train_path = tf.keras.utils.get_file(TRAIN_URL.split('\/')[-1], TRAIN_URL)\r\n    test_path = tf.keras.utils.get_file(TEST_URL.split('\/')[-1], TEST_URL)\r\n\r\n    return train_path, test_path\r\n\r\ndef load_data(y_name='Result'):\r\n    &quot;&quot;&quot;\u4ee5(train_x, train_y), (test_x, test_y)\u7684\u5f62\u5f0f\u8fd4\u56de\u6bd4\u8d5b\u6570\u636e\u96c6&quot;&quot;&quot;\r\n    train_path, test_path = maybe_download()\r\n\r\n    train = pd.read_csv(train_path, names=CSV_COLUMN_NAMES, header=0)\r\n    train_x, train_y = train, train.pop(y_name)\r\n\r\n    test = pd.read_csv(test_path, names=CSV_COLUMN_NAMES, header=0)\r\n    test_x, test_y = test, test.pop(y_name)\r\n\r\n    return (train_x, train_y), (test_x, test_y)\r\n\r\ndef train_input_fn(features, labels, batch_size):\r\n    &quot;&quot;&quot;\u8bad\u7ec3\u8f93\u5165\u51fd\u6570&quot;&quot;&quot;\r\n    # \u5c06\u8f93\u5165\u8f6c\u6362\u4e3a\u6570\u636e\u96c6\r\n    dataset = tf.data.Dataset.from_tensor_slices((dict(features), labels))\r\n\r\n    # \u6253\u4e71\u3001\u91cd\u590d\u4ee5\u53ca\u6346\u7ed1\u6837\u672c\r\n    dataset = dataset.shuffle(1000).repeat().batch(batch_size)\r\n\r\n    # \u8fd4\u56de\u6570\u636e\u96c6\r\n    return dataset\r\n\r\ndef eval_input_fn(features, labels, batch_size):\r\n    &quot;&quot;&quot;\u9a8c\u8bc1\u548c\u9884\u6d4b\u8f93\u5165\u51fd\u6570&quot;&quot;&quot;\r\n    features=dict(features)\r\n    if labels is None:\r\n        # \u6ca1\u6709\u6807\u7b7e\uff0c\u53ea\u6709\u7279\u5f81\r\n        inputs = features\r\n    else:\r\n        inputs = (features, labels)\r\n\r\n    # \u5c06\u8f93\u5165\u8f6c\u6362\u4e3a\u6570\u636e\u96c6\r\n    dataset = tf.data.Dataset.from_tensor_slices(inputs)\r\n\r\n    # \u6346\u7ed1\u6837\u672c\r\n    assert batch_size is not None, &quot;batch_size must not be None&quot;\r\n    dataset = dataset.batch(batch_size)\r\n\r\n    # \u8fd4\u56de\u6570\u636e\u96c6\r\n    return dataset\r\n\r\nCSV_TYPES = [[0.0], [0.0], [0.0], \r\n             [0.0], [0.0], [0,0],\r\n             [0.0], [0.0], [0.0],\r\n             [0.0], [0.0], [0.0],\r\n             [0.0], [0.0], [0.0],\r\n             [0.0], [0.0], [0]]\r\n\r\ndef _parse_line(line):\r\n    # \u89e3\u7801\u884c\r\n    fields = tf.decode_csv(line, record_defaults=CSV_TYPES)\r\n\r\n    # \u5c06\u7ed3\u679c\u6253\u5305\u4e3a\u5b57\u5178\r\n    features = dict(zip(CSV_COLUMN_NAMES, fields))\r\n\r\n    # \u5c06\u6807\u7b7e\u4ece\u7279\u5f81\u4e2d\u5206\u79bb\r\n    label = features.pop('Species')\r\n\r\n    return features, label\r\n\r\n\r\ndef csv_input_fn(csv_path, batch_size):\r\n    # \u521b\u5efa\u4e00\u4e2a\u5305\u542b\u591a\u884c\u6587\u672c\u7684\u6570\u636e\u96c6\r\n    dataset = tf.data.TextLineDataset(csv_path).skip(1)\r\n\r\n    # \u8f6c\u6362\u6bcf\u4e00\u884c\r\n    dataset = dataset.map(_parse_line)\r\n\r\n    # \u6253\u4e71\u3001\u91cd\u590d\u548c\u6346\u7ed1\u6837\u672c\r\n    dataset = dataset.shuffle(1000).repeat().batch(batch_size)\r\n\r\n    # \u8fd4\u56de\u6570\u636e\u96c6\r\n    return dataset\r\n<\/pre>\n<p>\u9996\u5148\u8bf4<em>WorldCupPredict.py<\/em>\u7684\u4ee3\u7801\u3002\u5f15\u7528\u4e86\u4e00\u4e9b\u56fa\u5b9a\u7684\u6a21\u5757\uff0c\u7136\u540e\u786e\u5b9a\u4e86\u6b65\u6570\uff08step\uff09\u548c\u6346\u7ed1\uff08batch\uff09\u4e24\u4e2a\u53c2\u6570\u3002\u6b65\u957f\u5e76\u4e0d\u662f\u8d8a\u5927\u8d8a\u597d\uff0c\u6a21\u578b\u7684\u51c6\u786e\u5ea6\u4f1a\u4ece\u7b2c\u4e00\u6b65\u5f00\u59cb\u9010\u6e10\u6ce2\u52a8\u4e0a\u5347\uff0c\u5230\u8fbe\u67d0\u4e2a\u6700\u5927\u503c\u503c\u540e\u53ef\u80fd\u4f1a\u4e0b\u964d\u4e00\u70b9\u7136\u540e\u518d\u4e0a\u5347\u4e00\u70b9\u3002\u5f53\u8bad\u7ec3\u6570\u636e\u591a\u7684\u65f6\u5019\uff0c\u6346\u7ed1\u503c\u53ef\u4ee5\u8bbe\u5927\u4e00\u70b9\u3002\u5177\u4f53\u610f\u4e49\u53c2\u8003TensorFlow\u5b98\u7f51\u7684\u672f\u8bed\u8868\u3002<\/p>\n<p>\u7136\u540e\u4f1a\u8c03\u7528game_data\u6a21\u5757\u83b7\u5f97\u8bad\u7ec3\u6570\u636e\u548c\u6d4b\u8bd5\u6570\u636e\u3002\u518d\u5bf9\u7279\u5f81\u503c\u8fdb\u884c\u6807\u8bb0\u3002\u4e4b\u540e\u6784\u5efa\u4e86\u4e00\u4e2a\u5206\u7c7b\u5668\uff0c\u56e0\u4e3a\u5b9e\u8d28\u4e0a\u8fd9\u662f\u4e2a\u6309\u7279\u5f81\u628a\u6bd4\u8d5b\u5f52\u7c7b\u4e3a\u4e3b\u80dc\u3001\u5e73\u5c40\u548c\u4e3b\u8d1f\u4e09\u79cd\u4e0d\u540c\u8d5b\u679c\u7684\u95ee\u9898\u3002<\/p>\n<p>\u5206\u7c7b\u5668\u4e2d\u8bbe\u7f6e\u4e86\u4e09\u4e2a\u9690\u85cf\u5c42\uff0c\u6bcf\u4e00\u4e2a\u7279\u5f81\u503c\u90fd\u4f1a\u6d41\u8fdb\u7b2c\u4e00\u5c42\u7684\u6bcf\u4e2a\u795e\u7ecf\u5143\u91cc\uff0c\u7b2c\u4e00\u5c42\u795e\u7ecf\u5143\u7684\u8ba1\u7b97\u7ed3\u679c\u53c8\u4f1a\u6d41\u8fdb\u7b2c\u4e8c\u5c42\uff0c\u5982\u6b64\u7c7b\u63a8\u3002\u76f4\u81f3\u8f93\u51fa\u5230\u4e09\u4e2a\u4e0d\u540c\u7684\u8d5b\u679c\u7c7b\u522b\u4e2d\u3002\u6700\u4f73\u7684\u9690\u85cf\u5c42\u6570\u7684\u8bbe\u7f6e\u548c\u6bcf\u5c42\u795e\u7ecf\u5143\u6570\u7684\u8bbe\u7f6e\u9700\u8981\u901a\u8fc7\u91cd\u590d\u5b9e\u9a8c\u548c\u4e00\u5b9a\u7684\u7ecf\u9a8c\u51b3\u5b9a\u3002\u5f80\u5f80\u591a\u7684\u5c42\u548c\u591a\u7684\u795e\u7ecf\u5143\u9700\u8981\u66f4\u591a\u7684\u8bad\u7ec3\u6570\u636e\u53bb\u8fbe\u5230\u6709\u6548\u7684\u8bad\u7ec3\u3002<\/p>\n<p>\u8bad\u7ec3\u5b8c\u6210\u540e\u4f1a\u4f7f\u7528\u6d4b\u8bd5\u6570\u636e\u5bf9\u6a21\u578b\u8fdb\u884c\u8bc4\u4f30\uff0c\u5e76\u7ed9\u51fa\u51c6\u786e\u5ea6\u3002\u4e4b\u540e\u53ef\u4ee5\u624b\u52a8\u6784\u9020\u4e00\u4e2a\u6216\u591a\u4e2a\u6bd4\u8d5b\u6570\u636e\u7ed9\u6a21\u578b\u9884\u6d4b\u3002\u8fd9\u91cc\u4f7f\u7528\u7684\u662f\u7b2c57\u573a\u4e4c\u62c9\u572d\u5bf9\u6218\u6cd5\u56fd\u7684\u6570\u636e\uff0c\u8fd9\u91cc\u5e76\u6ca1\u6709\u7ed9\u51fa\u8d5b\u679c\u4f5c\u4e3a\u6807\u7b7e\u503c\u3002<\/p>\n<p>\u6a21\u578b\u4f1a\u6839\u636e\u7279\u5f81\u503c\u9884\u6d4b\u51fa\u5f52\u5c5e\u4e8e\u4e0d\u540c\u6807\u7b7e\u7684\u53ef\u80fd\u6027\uff0c\u8fd9\u91cc\u6587\u672c\u8f93\u51fa\u4e86\u6700\u5927\u53ef\u80fd\u6027\u7684\u6807\u7b7e\u53ca\u5176\u53ef\u80fd\u6027\u3002<\/p>\n<p>\u63a5\u4e0b\u6765\u8bf4\u4e00\u4e0b<em>game_data.py<\/em>\u7684\u4ee3\u7801\u3002\u9996\u5148\u5b9a\u4e49\u4e86\u4e24\u4e2a\u6570\u636e\u6e90\uff0c\u91c7\u7528\u7684\u662f\u672c\u7f51\u7ad9\u6574\u7406\u597d\u7684\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u3002\u6570\u636e\u6e90\u7684\u5f62\u5f0f\u662fcsv\u6587\u4ef6\uff0c\u662f\u4e00\u79cd\u4ee5\u6362\u884c\u8868\u793a\u6570\u636e\u884c\uff0c\u9017\u53f7\u8868\u793a\u4e00\u884c\u4e2d\u7684\u6570\u636e\u5217\u7684\u8868\u683c\u578b\u6570\u636e\u3002<\/p>\n<p>\u9996\u5148\u9700\u8981\u4ece\u7f51\u7edc\u628a\u6570\u636e\u6e90\u4e0b\u8f7d\u56de\u5230\u672c\u5730\uff0c\u7136\u540e\u5c06csv\u6570\u636e\u6e90\u6574\u7406\u6210\u53ef\u4ee5\u88ab\u8ba1\u7b97\u7684\u5f20\u91cf\uff08tensor\uff09\u3002\u89e3\u6790csv\u6587\u4ef6\u9700\u8981\u7528\u5230\u4e00\u4e2a\u6570\u636e\u6a21\u677f\u5411\u7a0b\u5e8f\u8bf4\u660e\u6bcf\u5217\u662f\u4e00\u4e2a\u4ec0\u4e48\u6570\u636e\u7c7b\u578b\u7684\u6570\u636e\u3002\u8fd9\u91cc\u628a\u524d\u976217\u5217\u7684\u7279\u5f81\u503c\u4f5c\u4e3a\u6d6e\u70b9\u6570\uff0c\u6700\u540e\u4e00\u5217\u4f5c\u4e3a\u6574\u6570\u3002\u56e0\u4e3a\u5f20\u91cf\u8f6c\u6362\u8fc7\u7a0b\u4e2d\u6240\u6709\u6570\u636e\u7684\u7c7b\u578b\u8981\u76f8\u540c\u3002\u8fd9\u4e2a\u6587\u4ef6\u7684\u4ee3\u7801\u4e3b\u8981\u662f\u505a\u4e86\u6570\u636e\u6574\u7406\u7684\u5de5\u4f5c\u3002<\/p>\n<h3>\u8fd0\u884c\u7ed3\u679c<\/h2>\n<h3><a href="\""http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/TIM\u622a\u56fe20180706195409.png\"><img decoding=\"async\" loading=\"lazy\" class=\"alignnone wp-image-67 size-full\" src=\"http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/TIM\u622a\u56fe20180706195409.png\" alt=\"\" width=\"678\" height=\"223\" srcset=\"http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/TIM\u622a\u56fe20180706195409.png 678w, http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/TIM\u622a\u56fe20180706195409-300x99.png 300w\" sizes=\"(max-width: 678px) 100vw, 678px\" \/><\/a><\/h2>\n<p>\u8fc7\u7a0b\u4e2d\u4f1a\u7ed9\u51fa\u5728\u8bad\u7ec3\u6a21\u578b\u65f6\u7684\u4fe1\u606f\uff0c\u5305\u62ec\u4e86\u6b65\u6570\u3001\u8fdb\u884c\u5230\u5f53\u524d\u6b65\u6570\u7684\u635f\u5931\u548c\u6bcf\u79d2\u8fdb\u884c\u4e86\u591a\u5c11\u6b65\u3002<\/p>\n<p><a href="\""http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/TIM\u622a\u56fe20180706195427.png\"><img decoding=\"async\" loading=\"lazy\" class=\"alignnone wp-image-66 size-full\" src=\"http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/TIM\u622a\u56fe20180706195427.png\" alt=\"\" width=\"527\" height=\"224\" srcset=\"http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/TIM\u622a\u56fe20180706195427.png 527w, http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/TIM\u622a\u56fe20180706195427-300x128.png 300w\" sizes=\"(max-width: 527px) 100vw, 527px\" \/><\/a><\/p>\n<p>\u6a21\u578b\u8bad\u7ec3\u5b8c\u6210\u540e\uff0c\u4f1a\u653e\u5165\u6d4b\u8bd5\u96c6\uff0c\u7528\u4e8e\u8bc4\u4f30\u6a21\u578b\u7684\u51c6\u786e\u5ea6\u3002\u6700\u540e\u4f1a\u7ed9\u51fa\u4e00\u4e2a\u65e0\u6807\u7b7e\u7684\u7279\u5f81\u503c\u7684\u6807\u7b7e\u9884\u6d4b\u7ed3\u679c\u3002\u5982\u4e0a\u56fe\u6240\u793a\uff0c\u5373\u8bc4\u4f30\u51c6\u786e\u5ea6\u8fbe\u523050%\uff0c\u800c\u670939.5%\u7684\u6982\u7387\u672c\u573a\uff08\u4e4c\u62c9\u572dVS\u6cd5\u56fd\uff09\u7684\u6bd4\u8d5b\u4e2d\uff0c\u4e3b\u961f\uff08\u4e4c\u62c9\u572d\uff09\u4f1a\u8f93\u3002<\/p>\n<p>\u8bfb\u8005\u53ef\u4ee5\u5c1d\u8bd5\u8fdb\u884c\u4ee5\u4e0b\u51e0\u79cd\u5de5\u4f5c\uff1a<\/p>\n<ol>\n<li>\u8c03\u6574\u6b65\u957f\u548c\u6346\u7ed1\u5bf9\u8bad\u7ec3\u548c\u8bc4\u4f30\u7684\u5f71\u54cd\u3002<\/li>\n<li>\u8c03\u6574\u9690\u85cf\u5c42\u6570\u548c\u795e\u7ecf\u5143\u4e2a\u6570\u5bf9\u8bad\u7ec3\u548c\u8bc4\u4f30\u7684\u5f71\u54cd\u3002<\/li>\n<li>\u52a0\u5165\u65b0\u7684\u7279\u5f81\u5217\u548c\u5220\u9664\u5df2\u6709\u7279\u5f81\u5217\u5bf9\u8bad\u7ec3\u548c\u8bc4\u4f30\u7684\u5f71\u54cd\u3002<\/li>\n<li>\u628a\u8be5\u9879\u76ee\u4f5c\u4e3a\u6a21\u677f\u5904\u7406\u5176\u4ed6\u5206\u7c7b\u95ee\u9898\u3002<\/li>\n<li>\u58d5\u78382\u5757\u94b1\u6cd5\u56fd\u8d62\u3002<\/li>\n<\/ol>\n<h1>\u6570\u636e\u96c6\u6587\u4ef6<\/h1>\n<p><a href="\""http:\/\/www.halve.top\/wp-content\/uploads\/2018\/07\/\u4e16\u754c\u676f\u6570\u636e\u5e93.xlsx\">\u70b9\u6b64\u4e0b\u8f7d<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u76ee\u6807 \u672c\u6587\u7684\u76ee\u6807\u662f\u5728\u5b66\u4e60\u4e86TensorFlow\u7684\u5b98\u65b9Graph Execution\u793a\u4f8b\u540e\uff0c\u901a\u8fc7\u4e3e\u4e00\u53cd\u4e09\u8fd0\u7528\u5176\u4ed6 [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":72,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[21],"tags":[24,23,22],"_links":{"self":[{"href":"http:\/\/www.halve.top\/index.php?rest_route=\/wp\/v2\/posts\/62"}],"collection":[{"href":"http:\/\/www.halve.top\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.halve.top\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.halve.top\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.halve.top\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=62"}],"version-history":[{"count":8,"href":"http:\/\/www.halve.top\/index.php?rest_route=\/wp\/v2\/posts\/62\/revisions"}],"predecessor-version":[{"id":77,"href":"http:\/\/www.halve.top\/index.php?rest_route=\/wp\/v2\/posts\/62\/revisions\/77"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/www.halve.top\/index.php?rest_route=\/wp\/v2\/media\/72"}],"wp:attachment":[{"href":"http:\/\/www.halve.top\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=62"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.halve.top\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=62"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.halve.top\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=62"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}