{"id":3543,"date":"2022-07-14T16:06:16","date_gmt":"2022-07-14T08:06:16","guid":{"rendered":"https:\/\/aiwellbore.com\/?p=3543"},"modified":"2024-07-12T17:09:49","modified_gmt":"2024-07-12T09:09:49","slug":"%e7%9f%b3%e6%b2%b9%e4%b8%93%e4%b8%9a%ef%bc%9a%e4%b8%80%e6%96%87%e5%85%a5%e9%97%a8%e6%9c%ba%e5%99%a8%e5%ad%a6%e4%b9%a0%ef%bc%8c%e4%bb%a5%e6%b5%8b%e4%ba%95%e5%b2%a9%e6%80%a7%e5%88%86%e7%b1%bb%e9%a2%84","status":"publish","type":"post","link":"https:\/\/aiwellbore.com\/?p=3543","title":{"rendered":"\u77f3\u6cb9\u4e13\u4e1a\uff1a\u4e00\u6587\u5165\u95e8\u673a\u5668\u5b66\u4e60\uff0c\u4ee5\u6d4b\u4e95\u5ca9\u6027\u5206\u7c7b\u9884\u6d4b\u4e3a\u4f8b"},"content":{"rendered":"

https:\/\/blog.csdn.net\/weixin_45638544\/article\/details\/107429224<\/a>
\n@TOC<\/a><\/p>\n

1 \u524d\u8a00<\/h1>\n

1.1 \u673a\u5668\u5b66\u4e60\u7684\u76f8\u5173\u80cc\u666f<\/h2>\n

\n\u4ec5\u4e3a\u5206\u4eab\u77e5\u8bc6\uff0c\u5199\u4e0b\u672c\u6587\u3002\u5982\u6709\u9519\u8bef\uff0c\u8bf7\u89c1\u8c05\uff0c\u6b22\u8fce\u6279\u8bc4\uff01
\n<\/p>\n

\u968f\u7740\u6280\u672f\u7684\u8fed\u4ee3\uff0c\u4f20\u7edf\u6280\u672f\u5bf9\u4e8e\u77f3\u6cb9\u5de5\u4e1a\u7684\u652f\u6301\u6709\u7740\u4e00\u5b9a\u7684\u5236\u7ea6\uff0c\u57fa\u4e8e\u6570\u636e\u9a71\u52a8\u7684\u673a\u5668\u5b66\u4e60\u8fd9\u4e00\u6570\u5b66\u5de5\u5177\u80fd\u63d0\u4f9b\u66f4\u591a\u7684\u5e2e\u52a9\u3002\u6545\u6b64\u5199\u4e0b\u672c\u6587\uff0c\u4e0d\u80fd\u5b9e\u73b0\u6240\u6709\u7684\u7ec6\u8282\uff0c\u4f46\u662f\u53ef\u4ee5\u5c06\u6574\u4e2a\u673a\u5668\u5b66\u4e60\u7684\u601d\u60f3\u5927\u81f4\u4e88\u4ee5\u8bf4\u660e\u3002<\/p>\n

\u673a\u5668\u5b66\u4e60\u662f\u4e00\u4e2a\u6570\u5b66\u5de5\u5177\uff0c\u5305\u62ec\u4e86\u4f17\u591a\u7b97\u6cd5\uff0c\u4eba\u4eec\u719f\u77e5\u7684\u5373\u795e\u7ecf\u7f51\u7edc\uff08BP\uff09^1<\/a>\u3001\u652f\u6301\u5411\u91cf\u673a\uff08SVM\uff09^2<\/a>\u7b49\u7b49\u3002\u4ece\u5e94\u7528\u7684\u76ee\u7684\u4e0a\u53ef\u4ee5\u5206\u4e3a\u4e09\u79cd\uff0c\u5373\u76d1\u7763\u5b66\u4e60\u3001\u975e\u76d1\u7763\u5b66\u4e60\u3001\u5176\u4ed6\uff08\u5982\uff1a\u5f3a\u5316\u5b66\u4e60\uff09\u3002<\/p>\n

\n
graph LR\n\u673a\u5668\u5b66\u4e60 --> \u76d1\u7763\u5b66\u4e60 \n\u673a\u5668\u5b66\u4e60 --> \u975e\u76d1\u7763\u5b66\u4e60 \n\u673a\u5668\u5b66\u4e60 --> \u5f3a\u5316\u5b66\u4e60 \n\n\u76d1\u7763\u5b66\u4e60 --> \u56de\u5f52\n\u76d1\u7763\u5b66\u4e60 --> \u5206\u7c7b\n\n\u975e\u76d1\u7763\u5b66\u4e60  --> \u805a\u7c7b\n\u975e\u76d1\u7763\u5b66\u4e60 --> \u964d\u7ef4\n<\/code><\/pre>\n
\n
\u4e09\u79cd\u7c7b\u578b\u90fd\u6709\u5176\u5404\u81ea\u7684\u7279\u70b9\u53ca\u5e94\u7528\u573a\u666f\uff0c\u8fd9\u91cc\u5206\u522b\u53d9\u8ff0\u4e86\u5355\u7eaf\u7684\u4ece\u673a\u5668\u5b66\u4e60\u7684\u89d2\u5ea6\u51fa\u53d1\u4ee5\u53ca\u7ed3\u5408\u5230\u77f3\u6cb9\u5de5\u7a0b\u7684\u4e1a\u52a1\u573a\u666f\u5f53\u4e2d\u7684\u76f4\u89c2\u89e3\u91ca\u3002<\/p>\n\n\n\n\n\n\n\n
\u7c7b\u578b<\/th>\n\u673a\u5668\u5b66\u4e60\u7684\u7406\u8bba\u89d2\u5ea6<\/th>\n\u77f3\u6cb9\u5de5\u7a0b\u4e1a\u52a1\u89d2\u5ea6<\/th>\n<\/tr>\n<\/thead>\n
\u76d1\u7763\u5b66\u4e60<\/td>\n\u786e\u5b9a\u67d0\u4e2a\u72b6\u6001\u4e0b\u7684\u7b54\u6848<\/td>\n\u5982\uff1a\u4e00\u5b9a\u6761\u4ef6\u4e0b\u5ca9\u6027\u662f\u4ec0\u4e48<\/td>\n<\/tr>\n
\u975e\u76d1\u7763\u5b66\u4e60<\/td>\n\u6570\u636e\u5206\u5e03\u7684\u89c4\u5f8b<\/td>\n\u5982\uff1a \u4e24\u53e3\u4e95\u95f4\u662f\u5426\u5b58\u5728\u76f8\u4f3c\u6027<\/td>\n<\/tr>\n
\u5f3a\u5316\u5b66\u4e60<\/td>\n\u667a\u80fd\u4f53\u7684\u89c4\u5212\u95ee\u9898<\/td>\n\u5982\uff1a \u5982\u4f55\u5e03\u4e95\u4f7f\u5f97\u6548\u76ca\u6700\u4f73<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
\u5927\u591a\u6570\u7684\u8d44\u6599\u90fd\u96c6\u4e2d\u5728\u76d1\u7763\u5b66\u4e60\uff0c\u4ee5\u81f3\u4e8e\u5927\u5bb6\u5bf9\u4e8e\u673a\u5668\u5b66\u4e60\u7684\u8ba4\u77e5\u5b58\u5728\u504f\u5dee\uff0c\u4e2a\u4eba\u8ba4\u4e3a\u8d44\u6599\u96c6\u4e2d\u5728\u76d1\u7763\u5b66\u4e60\u7684\u539f\u56e0\u53ef\u4ee5\u5f52\u7ed3\u4e3a\uff1a\u4e0e\u5176\u4ed6\u76f8\u6bd4\uff0c\u76d1\u7763\u5b66\u4e60\u7684\u6548\u679c\u76f4\u89c2\uff0c\u5ba2\u89c2\u6027\u5f3a\uff0c\u4fbf\u4e8e\u8bc4\u4ef7\u3002<\/p>\n
1.2 \u673a\u5668\u5b66\u4e60\u7684\u516b\u80a1\u6587<\/h2>\n
\u673a\u5668\u5b66\u4e60\u65e2\u96be\u53c8\u7b80\u5355\uff0c\u96be\u7684\u5730\u65b9\u65e0\u9700\u591a\u8bf4\uff0c\u800c\u7b80\u5355\u7684\u5730\u65b9\u5c31\u5728\u4e8e\u673a\u5668\u5b66\u4e60\u662f\u6709\u56fa\u5b9a\u7684\u5957\u8def\u7684\uff0c\u4efb\u4f55\u7684\u673a\u5668\u5b66\u4e60\u90fd\u56f4\u7ed5\u7740\u4ee5\u4e0b\u4e94\u4e2a\u90e8\u5206\u8fdb\u884c\uff08\u7b2c\u56db\u6b65\u4e3b\u8981\u4e3a\u795e\u7ecf\u7f51\u7edc\u7684\u6a21\u578b<\/strong>\uff09\uff1a<\/p>\n
    \n
  1. \u6570\u636e\u5904\u7406\uff1a\u8bfb\u53d6\u6570\u636e   \u5904\u7406\u64cd\u4f5c<\/strong><\/li>\n
  2. \u6a21\u578b\u8bbe\u8ba1\uff1a\u7f51\u7edc\u7ed3\u6784\u8bbe\u8ba1<\/strong><\/li>\n
  3. \u8bad\u7ec3\u914d\u7f6e\uff1a\u4f18\u5316\u5668    \u8ba1\u7b97\u673a\u8d44\u6e90\u914d\u7f6e<\/strong><\/li>\n
  4. \u8bad\u7ec3\u8fc7\u7a0b\uff1a\u5faa\u73af\u8c03\u7528\u8bad\u7ec3\u8fc7\u7a0b   \u524d\u5411\u4f20\u64ad+\u635f\u5931\u51fd\u6570+\u53cd\u5411\u4f20\u64ad<\/strong><\/li>\n
  5. \u4fdd\u5b58\u6a21\u578b\uff1a\u5c06\u8bad\u7ec3\u597d\u7684\u6a21\u578b\u4fdd\u5b58<\/strong><\/li>\n<\/ol>\n
    \u65e2\u7136\u77e5\u9053\u4e86\u8fd9\u4e2a\u516b\u80a1\u6587\uff0c\u90a3\u4e48\u5b66\u4e60\u4e0e\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u89e3\u51b3\u5b9e\u9645\u95ee\u9898\u5c31\u7b80\u5355\u591a\u4e86\u3002<\/p>\n
    \u672c\u6848\u4f8b\u4e2d\u5c06\u6309\u7167\u8fd9\u4e2a\u987a\u5e8f\uff0c\u4ee5\u6d4b\u4e95\u5ca9\u6027\u5206\u7c7b\u4e3e\u4f8b\u8bf4\u660e\uff0c\u6570\u636e\u96c6\u5f15\u7528\u4e8eUniversity of Kansas Hugoton and Panoma gasfield<\/strong>^4<\/a><\/p>\n
    \u672c\u6848\u4f8b\u7a0d\u5fae\u5904\u7406\u7684\u6570\u636e\u540e\u7eed\u5e94\u8be5\u4f1a\u4e0a\u4f20\u81f3\u9644\u4ef6https:\/\/download.csdn.net\/download\/weixin_45638544\/12638256<\/a>\uff0c\u9644\u4ef6\u53ef\u80fd\u4f1a\u88ab\u5f3a\u5236\u8bbe\u4e3a\u6536\u8d39\uff0c\u5982\u679c\u6ca1\u627e\u5230\u6216\u4e0d\u65b9\u4fbf\u7684\u8bdd\u4e5f\u53ef\u4ee5\u5728\u8fd9\u4e2a\u8d85\u94fe\u63a5<\/a>\u514d\u8d39\u4e0b\u8f7d
    \n    https:\/\/gitee.com\/mmmahhhhe\/blog\/raw\/master\/other\/demo\/log_classifier\/log_data.xlsx<\/a>
    \n\u94fe\u63a5\u5728\u6d4f\u89c8\u5668\u6253\u5f00\u5e94\u8be5\u4f1a\u81ea\u52a8\u4e0b\u8f7d\u7684\uff0c\u5982\u679c\u6ca1\u4e0b\u8f7d\uff0c\u53ef\u4ee5\u6309ctrl+s\u624b\u52a8\u4e0b\u8f7d\uff0c\u522b\u8001\u95ee\u6211\u8981\u6570\u636e\u4e86q^q<\/strong><\/p>\n
    2 \u6a21\u578b\u5b9e\u6218<\/h1>\n
    \u672c\u6848\u4f8b\u6bd4\u8f83\u7b80\u5355\uff0c\u6545\u6b64\u6bcf\u4e2a\u5c0f\u8282\u6bd4\u8f83\u7b80\u77ed\uff0c\u51fa\u53d1\u70b9\u4ec5\u4e3a\u63d0\u4f9b\u4e00\u4e2a\u601d\u8def\uff0c\u5b9e\u9645\u6848\u4f8b\u53ef\u4ee5\u5c06\u6bcf\u4e2a\u90e8\u5206\u8fdb\u884c\u7279\u5b9a\u7684\u6269\u5c55\u3002<\/strong><\/p>\n
    2.0 \u5bfc\u5165\u76f8\u5173\u5e93<\/h2>\n
    \u672c\u6587\u91c7\u7528\u7684\u6570\u636e\u89c1\u9644\u4ef6<\/p>\n
    # \u672c\u6848\u4f8b\u5f15\u7528\u7684\u76f8\u5173\u5e93\nimport tensorflow as tf\nfrom tensorflow.keras import layers\nimport pandas as pd\nimport matplotlib.pyplot as plt<\/code><\/pre>\n
    # \u672c\u6848\u4f8b\u91c7\u7528\u7684TensorFlow2\u7248\u672c\nprint(tf.__version__)\nprint(tf.keras.__version__)\n2.1.0\n2.2.4-tf\n<\/code><\/pre>\n
    2.1 \u6570\u636e\u5904\u7406<\/h2>\n
    2.1.1 \u6570\u636e\u8bfb\u53d6<\/h3>\n
    # \u8bfb\u53d6\u539f\u59cb\u6570\u636e\nlog_data = pd.read_excel('log_data.xlsx')<\/code><\/pre>\n
    # \u6ce8\u610f\uff0c\u8be5\u4ee3\u7801\u4ee5ipynb\u4ea4\u4e92\u5f0f\u5448\u73b0\uff0c\u82e5\u4ee5py\u6587\u4ef6\u76f4\u63a5\u8fd0\u884c\u65f6\u53ef\u4ee5\u5c06\n# `log_data`\u4fee\u6539\u4e3aprint(log_data)  \n# \u4e0b\u6587\u540c\u7406\nlog_data\n\n    Depth   facno   facies  TH  U   K   RHOMAA  UMAA    PHIN\n0   13.5    1   Marine  7.653   9.172   1.209   2.776330    10.718131   42.099998\n1   14.0    1   Marine  7.794   9.209   1.210   2.741724    10.884575   43.200001\n2   14.5    1   Marine  7.567   9.129   1.159   2.698888    10.882272   40.400002\n3   15.0    1   Marine  7.296   9.010   1.095   2.698981    13.035485   37.799999\n4   15.5    1   Marine  9.105   9.554   1.715   2.710039    12.707702   35.500000\n... ... ... ... ... ... ... ... ... ...\n919 473.0   1   Marine  10.544  4.183   2.338   2.953034    8.994445    43.599998\n920 473.5   1   Marine  10.714  4.289   2.386   2.913379    9.434404    40.000000\n921 474.0   1   Marine  10.539  4.193   2.301   2.857787    9.567975    34.400002\n922 474.5   1   Marine  10.440  4.145   2.210   2.823199    9.857352    28.500000\n923 475.0   1   Marine  10.948  4.329   2.246   2.817633    9.586872    27.000000\n\n924 rows \u00d7 9 columns<\/code><\/pre>\n
    2.1.2 \u7f3a\u5931\u503c\u5904\u7406<\/h3>\n
    \u5176\u4e2dfacno<\/code> ,facies<\/code>\u5206\u522b\u4ee3\u8868\u5ca9\u6027\u7684\u7c7b\u522b\u5e8f\u53f7\u5373\u82f1\u6587\u540d\u3002facno<\/code> \u53d6\u503c\u4e3a0\u81f36\uff0c\u5176\u4e2d0\u4e3a\u672a\u77e5\u7684\u5ca9\u6027\u3002facies<\/code>\u4e2d\u672a\u77e5\u7684\u5ca9\u6027\u662f\u7f3a\u5931\u503c\uff0c\u6545\u6b64\u8fdb\u884c\u7f3a\u5931\u503c\u5904\u7406\u3002<\/p>\n
    # \u5c06 `facies`\u5217\u4e2d\u7a7a\u503c\u7f6e\u4e3a  Unknown\nlog_data['facies'][log_data['facies'].isnull()] = 'Unknown'<\/code><\/pre>\n
    2.1.3 \u6837\u672c\u7c7b\u522b\u5747\u8861<\/h3>\n
    \n
    \u5728\u4e0d\u540c\u6846\u67b6\u4e2d\u8fd9\u90e8\u5206\u4ee3\u7801\u9700\u8981\u5404\u81ea\u8c03\u6574\uff0c\u5177\u4f53\u8c03\u6574\u8d77\u6765\u6bd4\u8f83\u7e41\u7410\uff0c\u56e0\u6b64\u672c\u4f8b\u4e2d\u53ea\u662f\u8fdb\u884c\u4e86\u5747\u8861\u5206\u6790\uff0c\u5e76\u6ca1\u6709\u8fdb\u884c\u76f8\u5e94\u5904\u7406\u3002<\/p>\n<\/blockquote>\n
    \u8f93\u51fa\u5ca9\u6027\u7c7b\u578b\u53ca\u5bf9\u5e94\u5e8f\u53f7<\/p>\n
    # 6\u79cd\u5ca9\u6027\u7c7b\u578b\uff0c\u4ee5\u53ca\u672a\u77e5\u7c7b\u578b\n# \u6ce8\uff1a0\u5bf9\u5e94\u7684nan\u4ee3\u8868\u672a\u77e5 Unknown\n# \u5c06facies \u6309\u7167facno\u5206\u7ec4\uff0c\u50a8\u5b58\u5230\u5217\u8868\ngroup = log_data['facies'].groupby(log_data['facno'])\nclass_list = []\nfor id_num, name in group:\n    class_list.append([id_num, name.iloc[0]])\n<\/code><\/pre>\n
    class_list\n\n[[0, 'Unknown'],\n [1, 'Marine'],\n [2, 'Paralic'],\n [3, 'Floodplain'],\n [4, 'Channel'],\n [5, 'Splay'],\n [6, 'Paleosol']]<\/code><\/pre>\n
    \u5ca9\u6027\u7c7b\u522b\u7edf\u8ba1<\/p>\n
    # \u7edf\u8ba1\u5404\u4e2a\u5ca9\u6027\u7684\u6570\u91cf\u5206\u5e03\nlog_data['facies'].value_counts()\n\nChannel       262\nFloodplain    232\nParalic       156\nMarine        128\nPaleosol       66\nUnknown        42\nSplay          38\nName: facies, dtype: int64<\/code><\/pre>\n
    \u673a\u5668\u5b66\u4e60\u5b58\u5728\u5404\u79cd\u7ec6\u8282\u95ee\u9898\uff0c\u5f88\u662f\u56f0\u96be\uff0c\u5e78\u8fd0\u7684\u662f\u76f8\u4e92\u4e4b\u95f4\u4e0d\u662f\u5fc5\u8981\u6761\u4ef6\uff0c\u53ef\u6839\u636e\u81ea\u5df1\u7684\u5b9e\u9645\u9700\u6c42\u4e0e\u80fd\u529b\u8fdb\u884c\u8fdb\u4e00\u6b65\u7814\u7a76\u3002
    \n\u672c\u6848\u4f8b\u5b58\u5728\u4e00\u5b9a\u7684\u6837\u672c\u4e0d\u5747\u8861\u95ee\u9898\uff0c\u5f62\u8c61\u7684\u8bf4\u53ef\u4ee5\u7406\u89e3\u6210\uff1a
    \n\u5047\u5982\u6709100\u4e2a\u4eba\uff0c\u5176\u4e2dn\u4e2a\u597d\u4eba\uff0c\u5e0c\u671b\u4f60\u6784\u5efa\u4e00\u4e2a\u7cfb\u7edf\u6765\u8bc6\u522b\u51fa\u8c01\u662f\u597d\u4eba\u8c01\u662f\u574f\u4eba\uff0c\u5982\u679c\u5168\u731c\u597d\u4eba\uff0c\u5219\u6a21\u578b\u7684\u6b63\u786e\u7387\u4e3an%\u3002
    \n\u5173\u952e\u70b9\u5728\u4e8en\u5982\u679c\u4e3a50\u65f6\uff0c\u6a21\u578b\u6ca1\u6709\u4efb\u4f55\u4f5c\u7528\uff0c\u56e0\u4e3a\u6211\u6295\u786c\u5e01\u4e5f\u53ef\u4ee5\u731c\uff0c\u5982\u679cn\u4e3a99\uff0c\u5219\u6a21\u578b\u53ea\u9700\u8981\u5168\u90e8\u731c\u597d\u4eba\u5373\u53ef\uff0c\u51c6\u786e\u7387\u9ad8\u8fbe99%\uff0c\u53cd\u4e4b\u5982\u679cn\u4e3a1\uff0c\u5219\u5168\u731c\u597d\u4eba\u7684\u51c6\u786e\u7387\u4e3a1%\uff0c\u8fd9\u662f\u4e00\u4e2a\u53cd\u6a21\u578b\u3002
    \n\u5f53\u7136\uff0c\u6a21\u578b\u4e0d\u662f\u968f\u4fbf\u731c\u7684\uff0c\u6bd4\u598299%\u4e2a\u597d\u4eba\u5982\u679c\u5168\u731c\u597d\u4eba\uff0c\u5219\u51c6\u786e\u738799%\uff0c\u5982\u679c\u731c98\u4e2a\u597d\u4eba\uff0c\u6b63\u786e\u7387\u53cd\u800c\u4e0b\u964d\u4e86\uff0c\u60f3\u8fdb\u4e00\u6b65\u7814\u7a76\u53ef\u53c2\u8003AUG\u7b49\u6307\u6807^6<\/a>\u3002<\/p>\n
    \u6240\u4ee5\u95ee\u9898\u53ef\u4ee5\u8f6c\u5316\u4e3a\uff0c\u731c\u9519\u7684\u4ee3\u4ef7\u662f\u591a\u5c11\uff0c\u6bd4\u5982\u6211\u4eec\u8ba4\u4e3a\u731c\u9519\u4e00\u4e2a\u574f\u4eba\u7684\u4ee3\u4ef7\u4e3a0.9\uff0c\u731c\u9519\u4e00\u4e2a\u597d\u4eba\u7684\u4ee3\u4ef7\u6bd4\u8f83\u5c0f\u4ec5\u4e3a0.1\uff0c\u591a\u7c7b\u522b\u65f6\u4e5f\u5c31\u6784\u6210\u4e86\u4ee3\u4ef7\u77e9\u9635\uff0c\u4e5f\u53ef\u91c7\u7528\u5176\u4ed6\u65b9\u6cd5^7<\/a>\u3002<\/p>\n
    2.1.4 \u6837\u672c\u7c7b\u522b\u7f16\u7801\u5316<\/h3>\n\n\n\n\n\n\n\n
    \u6837\u672c\u7684\u7c7b\u522b\u4e3a\u5b57\u7b26\u4e32\uff0c\u8ba1\u7b97\u673a\u65e0\u6cd5\u76f4\u63a5\u5904\u7406\uff0c\u53ef\u4ee5\u5c06\u5176\u53d8\u6210\u5bf9\u5e94\u7684\u6570\u5b57\u3002\u4f46\u540c\u65f6\u5374\u4f1a\u51fa\u73b0\u53e6\u5916\u7684\u95ee\u9898\uff0c\u5373\u6837\u672c\u7c7b\u522b\u672c\u8eab\u4e0d\u5b58\u5728\u987a\u5e8f\uff0c\u5018\u82e5\u53d8\u6210\u6570\u5b57\u5219\u5b58\u5728\u5148\u540e\u7684\u987a\u5e8f\u5173\u7cfb\uff0c\u56e0\u6b64\u5f15\u5165\u72ec\u70ed\u7f16\u7801\u601d\u60f3\uff0c\u5c06n<\/code>\u4e2a\u6837\u672cc<\/code>\u4e2a\u7c7b\u522b\u7684\u6837\u672c\u6807\u7b7e\uff0c\u53d8\u4e3an*c<\/code>\u7684\u77e9\u9635
    \n\u4f8b\u59824<\/code>\u4e2a\u6837\u672c3<\/code>\u4e2a\u7c7b\u522b\uff1a
    \n\u539f\u59cb\u6807\u7b7e<\/th>\n    		Channel<\/th>\n<\/tr>\n<\/thead>\n
    Floodplain<\/td>\n<\/tr>\n
    Floodplain<\/td>\n<\/tr>\n
    Marine<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n\n\n\n\n\n\n
    \u72ec\u70ed\u7f16\u7801\u6807\u7b7e<\/th>\n1<\/th>\n0<\/th>\n0<\/th>\n<\/tr>\n<\/thead>\n
    0<\/td>\n1<\/td>\n0<\/td>\n<\/tr>\n
    0<\/td>\n1<\/td>\n0<\/td>\n<\/tr>\n
    0<\/td>\n0<\/td>\n1<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
    \u672c\u4f8b\u5171924<\/code>\u4e2a\u6837\u672c7<\/code>\u4e2a\u6807\u7b7e\uff08\u5305\u542b\u4f4d\u7f6e\u6807\u7b7e\uff09\uff0c\u6548\u679c\u5982\u4e0b\uff1a<\/p>\n
    # \u5c06\u9884\u6d4b\u503c\u53d8\u4e3a\u72ec\u70ed\u7f16\u7801\nlabel = pd.get_dummies(log_data['facno'])<\/code><\/pre>\n
    label\n\n    0   1   2   3   4   5   6\n0   0   1   0   0   0   0   0\n1   0   1   0   0   0   0   0\n2   0   1   0   0   0   0   0\n3   0   1   0   0   0   0   0\n4   0   1   0   0   0   0   0\n... ... ... ... ... ... ... ...\n919 0   1   0   0   0   0   0\n920 0   1   0   0   0   0   0\n921 0   1   0   0   0   0   0\n922 0   1   0   0   0   0   0\n923 0   1   0   0   0   0   0\n924 rows \u00d7 7 columns<\/code><\/pre>\n
    2.1.5 \u6570\u636e\u5408\u5e76\u4e0e\u7b5b\u9009<\/h3>\n
    \u8bfb\u5165\u6570\u636e\u7684facno<\/code> ,facies<\/code>\u5df2\u7ecf\u88ab\u6784\u5efa\u6210\u72ec\u70ed\u7f16\u7801\u4e86\uff0c\u6240\u4ee5\u53ef\u4ee5\u53bb\u9664\uff0c\u5e76\u4e14\u9700\u8981\u5c06\u72ec\u70ed\u7f16\u7801\u6807\u7b7e\u5408\u5e76\u5230\u6570\u636e\u7684\u53f3\u4fa7\uff0c\u5373\u5408\u5e76\u6210\u4e00\u4e2a\u77e9\u9635\u3002\u672c\u64cd\u4f5c\u662f\u56e0\u4e3a\u5728\u4e0b\u6587\u7684\u4e71\u5e8f\u4e2d\u907f\u514d\u6807\u7b7e\u4e0e\u6570\u636e\u4e0d\u5339\u914d\u800c\u8bbe\u5b9a\u7684\uff0c\u5b9e\u9645\u9879\u76ee\u4e2d\u4e5f\u9700\u8981\u8003\u8651\u672c\u64cd\u4f5c\uff0c\u4f46\u5177\u4f53\u5b9e\u73b0\u9700\u8981\u8c03\u6574\u3002<\/p>\n
    data_all = pd.concat([log_data, label], axis=1)\ndata = data_all.drop(['facno', 'facies'], axis=1)<\/code><\/pre>\n
    data\n\n    Depth   TH  U   K   RHOMAA  UMAA    PHIN    0   1   2   3   4   5   6\n0   13.5    7.653   9.172   1.209   2.776330    10.718131   42.099998   0   1   0   0   0   0   0\n1   14.0    7.794   9.209   1.210   2.741724    10.884575   43.200001   0   1   0   0   0   0   0\n2   14.5    7.567   9.129   1.159   2.698888    10.882272   40.400002   0   1   0   0   0   0   0\n3   15.0    7.296   9.010   1.095   2.698981    13.035485   37.799999   0   1   0   0   0   0   0\n4   15.5    9.105   9.554   1.715   2.710039    12.707702   35.500000   0   1   0   0   0   0   0\n... ... ... ... ... ... ... ... ... ... ... ... ... ... ...\n919 473.0   10.544  4.183   2.338   2.953034    8.994445    43.599998   0   1   0   0   0   0   0\n920 473.5   10.714  4.289   2.386   2.913379    9.434404    40.000000   0   1   0   0   0   0   0\n921 474.0   10.539  4.193   2.301   2.857787    9.567975    34.400002   0   1   0   0   0   0   0\n922 474.5   10.440  4.145   2.210   2.823199    9.857352    28.500000   0   1   0   0   0   0   0\n923 475.0   10.948  4.329   2.246   2.817633    9.586872    27.000000   0   1   0   0   0   0   0\n924 rows \u00d7 14 columns<\/code><\/pre>\n
    2.1.6 \u6570\u636e\u6807\u51c6\u5316<\/h3>\n
    \u6807\u51c6\u5316\u662f\u5c06\u6570\u636e\u4ece\u539f\u59cb\u7684\u5206\u5e03\u8303\u56f4\u7f29\u5c0f\u5230\u4e00\u5b9a\u7684\u533a\u95f4\u5185\uff0c\u4e00\u822c\u4e3a[0, 1]<\/code>\u3002
    \n\u4f8b\u5982\u6e29\u5ea6\u7684\u6570\u636e\u70b9\u4e3a100, 101, 102<\/code>\uff0c\u53ef\u4ee5\u53d8\u4e3a0, 0.5, 1<\/code><\/p>\n
    \u6807\u51c6\u5316\u4e3b\u8981\u53ef\u4ee5\u52a0\u5feb\u6c42\u89e3\u901f\u5ea6\uff0c\u5728\u8bf8\u5982\u51b3\u7b56\u6811\u7b49\u7b97\u6cd5\u4e2d\u4e0d\u9700\u8981\u4f7f\u7528^8<\/a>\u3002<\/p>\n
    # \u6807\u51c6\u5316\ndata = (data - data.min()) \/ (data.max() - data.min())<\/code><\/pre>\n
    data\n\n    Depth   TH  U   K   RHOMAA  UMAA    PHIN    0   1   2   3   4   5   6\n0   0.000000    0.272580    0.657763    0.460292    0.629186    0.711912    0.715247    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n1   0.001083    0.279283    0.660173    0.460687    0.571649    0.732604    0.739910    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n2   0.002167    0.268492    0.654961    0.440537    0.500429    0.732317    0.677130    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n3   0.003250    0.255609    0.647208    0.415251    0.500583    1.000000    0.618834    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n4   0.004334    0.341605    0.682650    0.660213    0.518968    0.959251    0.567265    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n... ... ... ... ... ... ... ... ... ... ... ... ... ... ...\n919 0.995666    0.410011    0.332725    0.906361    0.922978    0.497627    0.748879    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n920 0.996750    0.418093    0.339631    0.925326    0.857047    0.552322    0.668161    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n921 0.997833    0.409774    0.333377    0.891742    0.764619    0.568927    0.542601    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n922 0.998917    0.405067    0.330250    0.855788    0.707111    0.604902    0.410314    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n923 1.000000    0.429217    0.342237    0.870012    0.697857    0.571276    0.376682    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n924 rows \u00d7 14 columns<\/code><\/pre>\n
    2.1.7 \u6570\u636e\u4e71\u5e8f<\/h3>\n
    \u6570\u636e\u7684\u5582\u5165\u987a\u5e8f\u5bf9\u4e8e\u6a21\u578b\u7684\u6548\u679c\u5b58\u5728\u4e00\u5b9a\u7684\u5f71\u54cd\uff0c\u6a21\u578b\u5bf9\u4e8e\u540e\u5582\u5165\u7684\u6570\u636e\u4f1a\u6709\u4e00\u4e9b\u504f\u597d\uff0c\u4e5f\u5c31\u662f\u53ef\u80fd\u503e\u5411\u4e8e\u5f97\u51fa\u540e\u5582\u5165\u6570\u636e\u7684\u7ed3\u8bba^9<\/a>\u3002<\/p>\n
    data_shuffle = data.sample(frac=1).reset_index(drop=True)<\/code><\/pre>\n
    data_shuffle\n\nDepth   TH  U   K   RHOMAA  UMAA    PHIN    0   1   2   3   4   5   6\n0   0.083424    0.442147    0.245488    0.642829    0.828245    0.612717    0.800448    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n1   0.289274    0.272010    0.195192    0.155275    0.519398    0.080124    0.278027    0.0 0.0 0.0 1.0 0.0 0.0 0.0\n2   0.735645    0.323398    0.173562    0.538127    0.519404    0.275792    0.448431    0.0 0.0 0.0 0.0 0.0 1.0 0.0\n3   0.517876    0.086613    0.161183    0.220071    0.427236    0.063992    0.459641    0.0 0.0 0.0 0.0 1.0 0.0 0.0\n4   0.152763    0.000000    0.091798    0.183327    0.321834    0.036976    0.383408    1.0 0.0 0.0 0.0 0.0 0.0 0.0\n... ... ... ... ... ... ... ... ... ... ... ... ... ... ...\n919 0.468039    0.052957    0.090234    0.158040    0.383710    0.041824    0.457399    0.0 0.0 0.0 0.0 1.0 0.0 0.0\n920 0.738895    0.423940    0.180533    0.570525    0.629411    0.308377    0.491031    0.0 0.0 0.0 0.0 0.0 1.0 0.0\n921 0.968581    0.442860    0.196821    0.681944    0.709012    0.403376    0.266816    0.0 1.0 0.0 0.0 0.0 0.0 0.0\n922 0.334778    0.513928    0.303603    0.613591    0.801850    0.537676    0.484305    0.0 0.0 0.0 1.0 0.0 0.0 0.0\n923 0.812568    0.058756    0.121767    0.217701    0.440332    0.091793    0.446188    0.0 0.0 0.0 0.0 1.0 0.0 0.0\n924 rows \u00d7 14 columns<\/code><\/pre>\n
    2.1.8 \u6570\u636e\u5212\u5206<\/h3>\n
    \u6211\u4eec\u5c06\u5168\u90e8\u6837\u672c\u5212\u5206\u4e3a\u4e24\u4e2a\u90e8\u5206\uff0c\u6216\u4e09\u4e2a\u90e8\u5206\uff0c\u5206\u522b\u4e3a\u8bad\u7ec3\u96c6\u3001\u9a8c\u8bc1\u96c6\u3001\u6d4b\u8bd5\u96c6\uff0c\u6709\u7684\u65f6\u5019\u9a8c\u8bc1\u96c6\u4f1a\u88ab\u7701\u7565\u4e5f\u5c31\u9000\u5316\u6210\u4e24\u4e2a\u90e8\u5206\u3002\u5176\u5212\u5206\u6bd4\u4f8b\u4e00\u822c\u4e3a7:2:1\uff0c\u62167:3\uff0c\u62168:2\u3002
    \n\u8bad\u7ec3\u96c6\u76f8\u5f53\u4e8e\u8001\u5e08\u5e03\u7f6e\u7684\u4f5c\u4e1a\uff0c\u9a8c\u8bc1\u96c6\u6216\u6d4b\u8bd5\u96c6\u76f8\u5f53\u4e8e\u671f\u672b\u8003\u8bd5\u3002\u5177\u4f53\u6765\u770b\u5212\u5206\u9a8c\u8bc1\u96c6\u548c\u6d4b\u8bd5\u96c6\u7684\u76ee\u7684\u662f\u9700\u8981\u5728\u6211\u4eec\u5b9e\u9a8c\u4e2d\u7528\u8bad\u7ec3\u96c6\u8bad\u7ec3\u540e\uff0c\u4f7f\u7528\u9a8c\u8bc1\u96c6\u8fdb\u884c\u6a21\u578b\u53c2\u6570\u8c03\u6574\uff0c\u5f53\u6211\u4eec\u8ba4\u4e3a\u9a8c\u8bc1\u96c6\u7684\u6548\u679c\u8fbe\u5230\u6807\u51c6\u540e\uff0c\u5373\u53ef\u4f7f\u7528\u6d4b\u8bd5\u96c6\u8fdb\u884c\u4eff\u771f\u6d4b\u8bd5\uff0c\u5373\u6a21\u62df\u771f\u5b9e\u6761\u4ef6\u4e0b\u672a\u77e5\u6570\u636e\u7684\u9884\u6d4b\u6548\u679c    ^10<\/a>\uff0c\u5982\u679c\u662f\u5c0f\u578b\u9879\u76ee\u4e00\u822c\u4e5f\u53ef\u4ee5\u53ea\u5212\u5206\u4e3a\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\uff0c\u51cf\u5c11\u5de5\u4f5c\u91cf\u3002<\/p>\n
    \u672c\u6848\u4f8b\u91c7\u75287:2:1<\/code><\/p>\n
    # \u8bad\u7ec3\u96c6\u3001\u9a8c\u8bc1\u96c6\u3001\u6d4b\u8bd5\u96c6 7:2:1\nratio_val = int(len(data_shuffle) * 0.7)\nratio_test = int(len(data_shuffle) * 0.8)\n\nx_train = data_shuffle.iloc[: ratio_val, :-7].values\ny_train = data_shuffle.iloc[: ratio_val, -7:].values\n\nx_val = data_shuffle.iloc[ratio_val: ratio_test, :-7].values\ny_val = data_shuffle.iloc[ratio_val: ratio_test, -7:].values\n\nx_test = data_shuffle.iloc[ratio_test:, :-7].values\ny_test = data_shuffle.iloc[ratio_test:, -7:].values<\/code><\/pre>\n
    \u8bad\u7ec3\u96c6\u3001\u9a8c\u8bc1\u96c6\u3001\u6d4b\u8bd5\u96c6\u7684\u5212\u5206\u70b9\u5982\u4e0b\uff0c\u4e24\u4e2a\u70b9\u6784\u62103\u4e2a\u533a\u95f4\u3002<\/p>\n
    ratio_val, ratio_test\n\n(646, 739)<\/code><\/pre>\n
    x_train<\/code>\u4ee3\u8868\u8bad\u7ec3\u96c6\u7684\u8f93\u5165\u6570\u636e\uff0c.iloc[: ratio_val, :-7]<\/code>\u4e2d\u7684-7<\/code>\u662f\u4ee3\u8868\u540e7\u5217\u6570\u636e\u5747\u4e3a\u6807\u7b7ey<\/code>\uff0c.values<\/code>\u4ee3\u8868\u5c06pandas\u7684DataFrame\u6570\u636e\u683c\u5f0f\u8f6c\u4e3anumpy<\/code>\u7684array<\/code>\u683c\u5f0f\uff0c\u8fd9\u79cd\u683c\u5f0f\u8f6c\u6362\u662f\u4e3a\u4e86\u4e0b\u6587\u5c06\u6570\u636e\u5bfc\u5165\u6a21\u578b\u3002\u4e00\u822c\u6765\u8bb2numpy<\/code>\u7528\u5904\u8f83\u591a\uff0c\u672c\u4f8b\u6bd4\u8f83\u7b80\u5355\u5c31\u6ca1\u6d89\u53ca\uff0c\u6587\u7ae0\u5f00\u5934\u4e5f\u5e76\u6ca1\u6709\u8c03\u7528numpy<\/code>\u5e93\u3002x_train<\/code>\u5f62\u72b6\u4e3a(n * 7)<\/code>
    \ny_train<\/code>\u4ee3\u8868\u8bad\u7ec3\u96c6\u7684\u6807\u7b7e\uff0c\u5f62\u72b6\u4e3a(n * 7)<\/code>\uff0cx_train<\/code>\u4e0ey_train<\/code>\u7684\u7b2c\u4e8c\u4e2a\u7ef4\u5ea6\u90fd\u662f7\u53ea\u662f\u672c\u6587\u78b0\u5de7\u7684\uff0c\u4e00\u822c\u6765\u8bf4x\u7684\u7ef4\u5ea6\u4f1a\u5927\u4e8ey\u7684\u7ef4\u5ea6\u3002<\/p>\n
    # \u5206\u522b\u5b9a\u4e49\u6570\u636e\u7684\u8f93\u5165\u7ef4\u5ea6\u4e0e\u8f93\u51fa\u7ef4\u5ea6\uff0c\u672c\u4f8b\u6070\u597d\u90fd\u4e3a7\nx_num = x_train.shape[1]\ny_num = y_train.shape[1]<\/code><\/pre>\n
    x_num, y_num\n\n(7, 7)<\/code><\/pre>\n
    2.2 \u6a21\u578b\u8bbe\u8ba1<\/h2>\n
    TensorFlow<\/code>\uff08\u7b80\u79f0tf<\/code>\uff09\u6bd4\u8f83\u5e9e\u5927\uff0c\u7ec6\u8282\u8f83\u591a\uff0c\u5176\u903b\u8f91\u8f83\u96be\u7406\u89e3\u3002
    \n\u4e00\u822c\u6765\u8bb2\u5927\u591a\u6570\u4eba\u6bd4\u8f83\u4e60\u60ef\u50cfipynb<\/code>\u8fd9\u79cd\u5373\u65f6\u663e\u793a\u6548\u679c\u7684\u89e3\u91ca\u6027\u7f16\u7a0b\u65b9\u6848\uff0c\u4f46\u662ftf<\/code>\u662f\u9759\u6001\u56fe\uff0c\u53ef\u80fd\u8fd9\u4e2a\u6982\u5ff5\u6bd4\u8f83\u62bd\u8c61\uff0c\u901a\u4fd7\u6765\u8bf4\u5c31\u662f\u7f16\u7a0b\u7684\u65f6\u5019\u5148\u5b9a\u4e49\u5404\u79cd\u53d8\u91cf\uff0c\u5e76\u4e14\u5c06\u5404\u4e2a\u53d8\u91cf\u653e\u5728\u4e00\u8d77\u7ec4\u5408\u6210\u4e00\u4e2a\u5730\u56fe\uff0c\u5f53\u7a0b\u5e8f\u6267\u884c\u65f6\u50cf\u6c34\u4e00\u6837\uff0c\u4f9d\u6b21\u8d70\u8fc7\u5404\u4e2a\u5730\u56fe\u4e0a\u7684\u4f4d\u7f6e\u3002
    \n\u542c\u4e0a\u53bb\u6ca1\u4ec0\u4e48\uff0c\u5b9e\u9645\u7f16\u7a0b\u4e2d\u4f1a\u9047\u5230\u5404\u79cd\u9519\u8bef\u96be\u4ee5\u53d1\u73b0\u3002
    \ntf<\/code>\u5c06keras<\/code>\u8fdb\u884c\u4e86\u5c01\u88c5\uff0ckeras<\/code>\u7f16\u7a0b\u76f8\u5bf9\u7b80\u5355\u70b9\uff0c\u5982\u679c\u7528\u7eaftf<\/code>\u6846\u67b6\u505a\u8d77\u6765\u4f1a\u6bd4\u8f83\u9ebb\u70e6\u3002\u672c\u4f8b\u91c7\u7528tensorflow.keras<\/code>\u8fd9\u4e00api\u8fdb\u884c\u5efa\u7acb\u3002<\/p>\n
    model = tf.keras.Sequential()\nmodel.add(layers.Dense(32, activation='relu'))\nmodel.add(layers.Dense(32, activation='relu'))\nmodel.add(layers.Dense(y_num, activation='softmax'))<\/code><\/pre>\n
    layers<\/code>\u7ee7\u627f\u4e8etensorflow.keras<\/code>\uff0c\u4e0b\u6587\u7f29\u5199\u4e3alayers<\/code>.
    \nlayers.Dense()<\/code>\u8fd9\u4e00api\u76f8\u5f53\u4e8e\u5efa\u7acb\u4e00\u4e2a\u5168\u8fde\u63a5\u5c42\uff1b
    \n32<\/code>\u5373\u4e3a\u8be5\u5c42\u8f93\u51fa\u7684\u7ef4\u5ea6\u662f32\uff0c\u4e00\u822c\u6765\u8bb2\uff08\u5c24\u5176\u662f\u56fe\u50cf\u68c0\u6d4b\u4efb\u52a1\uff09\u7ef4\u5ea6\u6570\u90fd\u662f2\u7684\u500d\u6570\u5e76\u4e14\u6bcf\u5c42\u7684\u7ef4\u5ea6\u6570\u7ffb\u500d\uff0c\u6bd4\u598232,64,128,256<\/code>\u7b49
    \nactivation<\/code>\u5373\u4e3a\u6307\u7ecf\u8fc7\u8be5\u5c42\u8ba1\u7b97\u540e\u8fdb\u884c\u6fc0\u6d3b\u8ba1\u7b97\uff0c\u4e00\u822c\u6765\u8bf4\u6fc0\u6d3b\u51fd\u6570\u90fd\u91c7\u7528relu<\/code>\uff0c\u53ea\u6709\u7f51\u7edc\u7684\u6700\u540e\u4e00\u5c42\u91c7\u7528sigmoid<\/code>\u6216softmax<\/code>\uff0c\u524d\u8005\u7528\u4e8e\u56de\u5f52\uff0c\u540e\u8005\u7528\u4e8e\u5206\u7c7b\uff0c\u672c\u4f8b\u662f\u5206\u7c7b\uff0c\u6545\u6b64\u91c7\u7528\u540e\u8005    ^11<\/a>\u3002<\/p>\n
    \u4e0b\u9762\u5f00\u59cb\u8bad\u7ec3\u914d\u7f6e\u3001\u8bad\u7ec3\u8fc7\u7a0b\u3001\u4fdd\u5b58\u6a21\u578b\uff08\u52a0\u8f7d\u6a21\u578b\uff09\u4e00\u822c\u7684\u5957\u8def\u5982\u4e0b\uff1a<\/p>\n
    2.3 \u8bad\u7ec3\u914d\u7f6e<\/h2>\n
    model.compile(optimizer=tf.keras.optimizers.Adam(0.001),\n             loss=tf.keras.losses.categorical_crossentropy,\n             metrics=[tf.keras.metrics.categorical_accuracy])<\/code><\/pre>\n
    \u6a21\u578b\u7684\u914d\u7f6e\u53ea\u6709\u4e00\u884c\u4ee3\u7801\uff0c\u5176\u4e2d\u5404\u4e2a\u7ec6\u8282\uff0c\u6bd4\u5982\u4f18\u5316\u5668\uff0c\u635f\u5931\u51fd\u6570\u90fd\u53ef\u4ee5\u8fdb\u884c\u6269\u5145\u6216\u81ea\u5b9a\u4e49\uff0c\u5927\u591a\u6570\u60c5\u51b5\u6846\u67b6\u81ea\u5e26\u7684\u4f18\u5316\u5668\u5373\u53ef\u3002
    \nmodel.compile<\/code>\u5c06\u524d\u9762\u5b9a\u4e49\u7684\u6a21\u578b\u8fdb\u884c\u7f16\u8bd1\uff0c\u5373\u5efa\u7acb\u9759\u6001\u56fe\uff1b
    \noptimizer<\/code>\u5b9a\u4e49\u4f18\u5316\u5668\uff0c\u5b98\u65b9\u6587\u6863\u63d0\u4f9b\u4e86\u81ea\u5e26\u7684\u4f18\u5316\u5668\uff1b
    \nloss<\/code>\u5b9a\u4e49\u635f\u5931\u51fd\u6570\uff0c\u7528\u4e8e\u4f18\u5316\u53c2\u6570\uff1b
    \nmetrics<\/code>\u5b9a\u4e49\u8bc4\u4ef7\u6807\u51c6\uff0ccategorical_accuracy<\/code>\u662f\u5206\u7c7b\u95ee\u9898\u7684\u901a\u7528\u65b9\u6cd5    ^12<\/a>\u3002<\/p>\n
    2.4 \u8bad\u7ec3\u8fc7\u7a0b<\/h2>\n
    2.4.1 \u8bad\u7ec3<\/h3>\n
    \u8bad\u7ec3\u8fc7\u7a0b\u4e3a\u6846\u67b6\u6240\u5c01\u88c5\uff0c\u4e5f\u6bd4\u8f83\u56fa\u5b9a\uff0c\u4f20\u5165\u6307\u5b9a\u53c2\u6570\u5373\u53ef<\/p>\n
    model.fit(x_train, y_train,\n          epochs=100, batch_size=100,\n          validation_data=(x_val, y_val))<\/code><\/pre>\n
    x_train<\/code>\u662f\u8bad\u7ec3\u96c6\u7684\u8f93\u5165\u6570\u636e
    \ny_train<\/code>\u662f\u8bad\u7ec3\u96c6\u7684\u6807\u7b7e
    \nvalidation_data<\/code>\u4f20\u5165\u5143\u7ec4\u662f\u9a8c\u8bc1\u96c6\u7684\u8f93\u5165\u4e0e\u6807\u7b7e
    \nepochs<\/code>\u8bad\u7ec3\u8f6e\u6b21\uff0c\u6570\u503c\u8d8a\u5927\u6548\u679c\u8d8a\u597d\uff0c\u540c\u65f6\u82b1\u8d39\u65f6\u95f4\u8d8a\u957f
    \nbatch_size<\/code>\u6bcf\u6b21\u5582\u5165batch_size<\/code>\u4e2a\u6837\u672c\uff0c\u4e00\u822c\u6765\u8bf4\u6839\u636e\u8ba1\u7b97\u673a\u7684\u914d\u7f6e\u6765\u786e\u5b9a\uff0c\u5c24\u5176\u662f\u5728\u56fe\u50cf\u95ee\u9898\u65b9\u9762\uff0c\u7531\u4e8e\u56fe\u50cf\u5360\u7528\u5185\u5b58\u8f83\u5927\uff0cbatch_size<\/code>\u8d8a\u5927\u5bf9\u4e8e\u8ba1\u7b97\u673a\u7684\u8981\u6c42\u8d8a\u9ad8\uff0c\u4f46\u662f\u672c\u6848\u4f8b\u8fd9\u7c7b\u578b\u7684\u6570\u636e\u6bd4\u8f83\u5c0f\uff0c\u53ef\u4ee5\u8bbe\u7684\u5927\u4e00\u70b9\u3002<\/p>\n
    \u8fd9\u884c\u4ee3\u7801\u8fd0\u7b97\u662f\u6700\u8017\u8d39\u65f6\u95f4\u7684\uff08\u672c\u4f8b\u6bd4\u8f83\u5c0f\u6240\u4ee5\u7528\u65f6\u77ed\uff09\uff0c\u53ef\u4ee5\u91c7\u7528GPU\u7b49\u52a0\u901f^13<\/a>\u3002<\/p>\n
    \u4e00\u5171\u8fd0\u884c100\u8f6e\uff0c\u524d3\u8f6e\u548c\u540e3\u8f6e\u7684\u65e5\u5fd7\u5982\u4e0b\uff1a<\/p>\n
    Train on 646 samples, validate on 93 samples\nEpoch 1\/100\n646\/646 [==============================] - 0s 32us\/sample - loss: 0.2418 - categorical_accuracy: 0.9226 - val_loss: 0.3763 - val_categorical_accuracy: 0.8710\nEpoch 2\/100\n646\/646 [==============================] - 0s 29us\/sample - loss: 0.2407 - categorical_accuracy: 0.9272 - val_loss: 0.3761 - val_categorical_accuracy: 0.8710\nEpoch 3\/100\n646\/646 [==============================] - 0s 28us\/sample - loss: 0.2408 - categorical_accuracy: 0.9272 - val_loss: 0.3827 - val_categorical_accuracy: 0.8710<\/code><\/pre>\n
    Epoch 97\/100\n646\/646 [==============================] - 0s 28us\/sample - loss: 0.2070 - categorical_accuracy: 0.9381 - val_loss: 0.3564 - val_categorical_accuracy: 0.8710\nEpoch 98\/100\n646\/646 [==============================] - 0s 26us\/sample - loss: 0.2082 - categorical_accuracy: 0.9396 - val_loss: 0.3585 - val_categorical_accuracy: 0.8710\nEpoch 99\/100\n646\/646 [==============================] - 0s 28us\/sample - loss: 0.2069 - categorical_accuracy: 0.9412 - val_loss: 0.3607 - val_categorical_accuracy: 0.8710\nEpoch 100\/100\n646\/646 [==============================] - 0s 29us\/sample - loss: 0.2072 - categorical_accuracy: 0.9427 - val_loss: 0.3605 - val_categorical_accuracy: 0.8710<\/code><\/pre>\n
    2.4.2 \u6d4b\u8bd5<\/h3>\n
    \u8fd9\u91cc\u5c06\u6d4b\u8bd5\u6570\u636e\u8fdb\u884c\u9884\u6d4b\uff0c\u5f97\u5230\u5176\u635f\u59310.3<\/code>\u548c\u5206\u7c7b\u51c6\u786e\u738789%<\/code><\/p>\n
    test_loss, test_acc = model.evaluate(x_test, y_test, batch_size=32)<\/code><\/pre>\n
    test_loss, test_acc\n\n(0.33426960110664367, 0.8918919)<\/code><\/pre>\n
    \u5404\u4e2a\u6837\u672c\u7684\u9884\u6d4b\u7ed3\u679c\u5982\u4e0b<\/p>\n
    predict = model.predict(x_test)<\/code><\/pre>\n
    \u4f46\u662f\u8fd9\u91cc\u8f93\u51fa\u7684\u662f[n, 7]<\/code>\u7684\u77e9\u9635\uff0c\u56e0\u4e3a\u8f93\u51fa\u7684\u662f\u6bcf\u4e2a\u7c7b\u522b\u7684\u6982\u7387
    \n\u4ee5\u524d\u4e24\u4e2a\u6837\u672c\u7684\u6570\u636e\u505a\u505a\u6f14\u793a<\/p>\n
    predict[:2]\n\narray([[3.6217916e-01, 1.2751932e-09, 5.8891876e-03, 3.3476558e-01,\n        2.7861273e-01, 7.1206079e-05, 1.8482126e-02],\n       [3.1979426e-03, 3.7681293e-05, 9.7117370e-01, 1.6478648e-02,\n        6.4800410e-03, 3.0302584e-05, 2.6016452e-03]], dtype=float32)<\/code><\/pre>\n
    \u56e0\u6b64\u9700\u8981\u5c06\u5176\u7b5b\u9009\u51fa\u7c7b\u522b\u6982\u7387\u6700\u5927\u7684\u6240\u5c5e\u7c7b\u522b\uff0c\u5373\u4e3a\u9884\u6d4b\u7c7b\u522b<\/p>\n
    predict.argmax(axis=1)\n\narray([0, 2, 4, 3, 6, 1, 4, 4, 4, 2, 6, 4, 2, 4, 4, 1, 4, 4, 4, 2, 1, 1,\n       4, 3, 3, 6, 4, 4, 3, 4, 4, 4, 2, 4, 2, 3, 4, 1, 2, 1, 4, 1, 3, 3,\n       6, 3, 3, 3, 4, 3, 1, 3, 1, 0, 1, 6, 1, 4, 3, 3, 4, 3, 4, 6, 6, 3,\n       5, 2, 3, 5, 2, 6, 2, 4, 4, 1, 2, 2, 3, 3, 6, 3, 2, 2, 1, 6, 2, 2,\n       2, 2, 4, 2, 3, 2, 4, 3, 4, 2, 3, 4, 3, 2, 3, 2, 4, 6, 3, 2, 2, 4,\n       3, 6, 6, 3, 6, 3, 4, 4, 3, 1, 2, 3, 0, 3, 4, 1, 3, 3, 2, 2, 4, 1,\n       3, 1, 4, 2, 4, 4, 1, 4, 3, 5, 6, 5, 3, 3, 4, 4, 2, 1, 1, 2, 4, 3,\n       2, 1, 2, 6, 3, 1, 4, 2, 4, 2, 4, 3, 4, 5, 4, 1, 6, 1, 4, 2, 4, 4,\n       2, 4, 4, 4, 4, 3, 1, 3, 4], dtype=int64)<\/code><\/pre>\n
    \u53ef\u4ee5\u770b\u51fa\u6a21\u578b\u5bf9\u4e8e\u7b2c\u4e00\u4e2a\u6837\u672c\u7684\u9884\u6d4b\u7ed3\u679c\u4e3a0\uff0c\u524d\u6587\u5728\u6570\u636e\u5904\u7406\u65f6\u5df2\u77e5\u4e86\u7c7b\u522b\u4e3a0\u5bf9\u5e94\u7684\u771f\u5b9e\u6807\u7b7e\u4e3aUnknown<\/code>
    \n\u63a5\u4e0b\u6765\u770b\u4e00\u4e0b\u5b9e\u9645\u4e0a\u8fd9\u4e2a\u6837\u672c\u7684\u6807\u7b7e\u662f\u4ec0\u4e48<\/p>\n
    y_test.argmax(axis=1)\n\narray([6, 2, 4, 5, 3, 1, 4, 4, 4, 2, 6, 4, 2, 4, 4, 1, 4, 4, 4, 2, 1, 1,\n       4, 3, 3, 6, 4, 4, 3, 4, 4, 4, 2, 4, 1, 3, 4, 1, 2, 1, 4, 1, 3, 3,\n       6, 3, 3, 0, 0, 3, 1, 3, 1, 0, 1, 6, 1, 4, 3, 3, 4, 3, 4, 6, 6, 3,\n       3, 2, 3, 5, 2, 2, 2, 4, 4, 1, 2, 2, 3, 3, 6, 3, 2, 2, 1, 6, 2, 3,\n       2, 0, 4, 2, 3, 1, 4, 3, 4, 2, 3, 4, 3, 2, 2, 1, 3, 6, 3, 2, 2, 4,\n       3, 3, 6, 3, 6, 0, 4, 4, 3, 1, 2, 3, 0, 4, 4, 1, 3, 3, 2, 2, 4, 1,\n       3, 1, 4, 2, 4, 4, 1, 4, 3, 5, 6, 5, 3, 3, 4, 4, 2, 1, 1, 2, 4, 3,\n       2, 1, 2, 2, 3, 1, 4, 2, 4, 2, 4, 3, 4, 5, 4, 1, 3, 1, 4, 2, 4, 4,\n       2, 4, 4, 4, 4, 5, 1, 3, 4], dtype=int64)<\/code><\/pre>\n
    \u663e\u7136\u53ef\u4ee5\u77e5\u9053\u771f\u5b9e\u6807\u7b7e\u4e3a6<\/code>\u4e5f\u5c31\u662fPaleosol<\/code>\u9884\u6d4b\u9519\u8bef\uff0c<\/p>\n
    \u867d\u7136\u5df2\u7ecf\u77e5\u9053\u4e86\u51c6\u786e\u738789%\uff0c\u4e5f\u53ef\u4ee5\u5b9a\u4e49\u4e2a\u51fd\u6570\u76f4\u89c2\u7684\u611f\u53d7\u4e00\u4e0b\u9884\u6d4b\u503c\u4e0e\u771f\u5b9e\u503c\u7684\u5dee\u522b\u3002<\/p>\n
    def draw(*datas):\n    print(type(datas))\n    for data in datas:\n        plt.scatter(range(len(data)), data)\n    plt.show()\n\ndraw(y_test.argmax(axis=1) - predict.argmax(axis=1))<\/code><\/pre>\n
    \"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"
    \n\u7eb5\u5750\u6807\u7684\u503c\u5176\u5b9e\u6ca1\u4ec0\u4e48\u7269\u7406\u542b\u4e49\uff0c\u503c\u4e3a0\u7684\u5c31\u8868\u793a\u9884\u6d4b\u6b63\u786e<\/p>\n
    \u4e0b\u9762\u8fd9\u884c\u4ee3\u7801\u662f\u4e0a\u9762\u7684\u4ee3\u7801\u4e2d\u628a\u51cf\u53f7\u6539\u4e3a\u9017\u53f7\u4e86\uff0c\u5176\u5b9e\u672c\u6765\u662f\u8fd9\u6837\u5199\u7684\uff0c\u53ea\u662f\u770b\u4e0a\u53bb\u6709\u4e9b\u4e71\uff0c\u5c31\u505a\u5dee\u76f8\u51cf\u4e86\uff0c\u5927\u5bb6\u5728\u5b9e\u9645\u64cd\u4f5c\u65f6\u53ef\u4ee5\u628a\u8fd9\u4e2a\u51fd\u6570\u8f93\u5165\u7684\u6570\u636e\u8303\u56f4\u81ea\u884c\u8c03\u6574\uff0c\u5373\u53ef\u663e\u793a\u5176\u4e2d\u67d0\u4e2a\u90e8\u5206\u7684\u6570\u636e<\/p>\n
    draw(y_test.argmax(axis=1), predict.argmax(axis=1))<\/code><\/pre>\n
    \"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"
    \n\u663e\u793a\u51fa\u6765\u7684\u84dd\u8272\u70b9\u662f\u9884\u6d4b\u6570\u636e\u9884\u6d4b\u9519\u7684\u60c5\u51b5\uff0c\u9884\u6d4b\u6b63\u786e\u7684\u84dd\u8272\u70b9\u88ab\u6a58\u9ec4\u8272\u70b9\u906e\u6321\u4f4f\u4e86
    \n\u672c\u4f8b\u7684\u9884\u6d4b\u6b63\u786e\u7387\u4e3a89%\uff0c\u5b58\u5728\u8f83\u5927\u7684\u5f00\u53d1\u6f5c\u529b\uff0c\u5927\u5bb6\u5728\u5b9e\u9645\u9879\u76ee\u4e2d\u53ef\u4ee5\u91c7\u7528\u5404\u79cd\u6280\u5de7\u8fdb\u884c\u63d0\u5347^14<\/a>\u3002<\/p>\n
    \n
    \u6570\u636e\u51b3\u5b9a\u4e86\u673a\u5668\u5b66\u4e60\u7684\u4e0a\u9650\uff0c\u800c\u6211\u4eec\u53ea\u662f\u903c\u8fd1\u8fd9\u4e2a\u4e0a\u9650\u3002<\/p>\n<\/blockquote>\n
    2.5 \u4fdd\u5b58\u6a21\u578b<\/h2>\n
    \u6a21\u578b\u4fdd\u5b58\u4e00\u822c\u662f\u4e24\u79cd\u60c5\u51b5\uff0c\u8bad\u7ec3\u5b8c\u4e86\u8981\u9884\u6d4b\uff0c\u56e0\u4e3a\u5b9e\u9645\u9700\u6c42\u4e2d\u8bad\u7ec3\u4e0e\u9884\u6d4b\u662f\u5206\u79bb \uff0c\u5373\u9700\u8981\u90e8\u7f72\u7aef\u52a0\u8f7d\u5df2\u7ecf\u4fdd\u5b58\u7684\u7f51\u7edc\u53ca\u53c2\u6570\u3002
    \n\u53e6\u4e00\u79cd\u60c5\u51b5\u5373\u8bad\u7ec3\u65f6\u95f4\u8f83\u957f\uff0c\u9700\u8981\u4e34\u65f6\u4e2d\u65ad\uff0c\u6216\u8005\u8bad\u7ec3\u4e86\u4e00\u5b9a\u8f6e\u6b21\u540e\u8ba4\u4e3a\u6548\u679c\u4e0d\u7406\u60f3\u9700\u8981\u7ee7\u7eed\u8bad\u7ec3\uff0c\u5426\u5219\u5982\u679c\u4ece\u5934\u5f00\u59cb\u8bad\u7ec3\u5219\u6d6a\u8d39\u8d44\u6e90\u3002<\/p>\n
    model.save('all_model.h5')<\/code><\/pre>\n
    \u8fd9\u884c\u4ee3\u7801\u5c06\u6a21\u578b\u4fdd\u5b58\u4e3aall_model.h5<\/code>\u4e8c\u8fdb\u5236\u6587\u4ef6\uff0c\u4fbf\u4e8e\u540e\u7eed\u4f7f\u7528\u3002
    \n\u800c\u53cd\u4e4b\u8c03\u7528\u8be5\u6a21\u578b\u7684\u4ee3\u7801\u5982\u4e0b\uff0c\u5373\u53ef\u4ece\u4e8c\u8fdb\u5236\u6587\u4ef6\u4e2d\u8bfb\u53d6\u7f51\u7edc\u5230\u5185\u5b58\u7528\u4e8e\u540e\u7eed\u8ba1\u7b97\u3002<\/p>\n
    model = tf.keras.models.load_model('all_model.h5')<\/code><\/pre>\n
    \n
    3 \u653e\u5728\u6700\u540e\u7684\u8bdd<\/h1>\n
    \u5b9e\u9645\u4e0a\u7f51\u7edc\u90e8\u5206\u624d\u662f\u673a\u5668\u5b66\u4e60\uff08\u6216\u8005\u8bf4\u6df1\u5ea6\u5b66\u4e60\uff09\u7684\u91cd\u5934\u620f\uff0c\u5404\u79cd\u7f51\u7edc\u7ed3\u6784\u5341\u5206\u590d\u6742\uff0c\u4e0d\u8fc7\u5bf9\u4e8e\u795e\u7ecf\u7f51\u7edc\u8fd9\u4e00\u7b97\u6cd5\u57fa\u672c\u4e0a\u53ea\u6709\u5168\u8fde\u63a5\u548c\u5377\u79ef\u8fd9\u4e24\u4e2a\u4e3b\u529b\uff0c\u5f53\u7136\u5305\u62ec\u5176\u4ed6\u7684\u4e00\u4e9b\u64cd\u4f5c\uff0c\u4e0d\u8fc7\u6838\u5fc3\u8fd8\u662f\u8fd9\u4e8c\u8005\u3002
    \n\u5bf9\u4e8e\u5927\u591a\u6570\u5de5\u7a0b\u4eba\u5458\u6765\u8bf4\uff0c\u4f7f\u7528\u5df2\u7ecf\u7ecf\u8fc7\u957f\u671f\u9524\u70bc\u7684\u6210\u719f\u7f51\u7edc\u5373\u53ef\u5b8c\u6210\u7edd\u5927\u591a\u6570\u4efb\u52a1\uff0c\u5982\u679c\u6709\u80fd\u529b\u8fdb\u884c\u6539\u8fdb\uff0c\u5f53\u7136\u6548\u679c\u5f88\u597d\u3002<\/p>\n
    \u7f51\u7edc\u8bbe\u8ba1\u5b8c\u6210\u540e\u7684\u90e8\u5206\u4e5f\u662f\u5f88\u91cd\u8981\uff0c\u4e0d\u8fc7\u7edd\u5927\u591a\u6570\u9879\u76ee\u5176\u5b9e\u6ca1\u5fc5\u8981\u7ea0\u7ed3\u540e\u9762\u7684\u6b65\u9aa4\u4e86\uff0c\u6539\u8fdb\u70b9\u4e5f\u96c6\u4e2d\u5728loss<\/code>\u51fd\u6570\u5982\u4f55\u8bbe\u8ba1\uff0c\u5982\u4f55\u964d\u4f4eloss<\/code>\u503c\uff0c\u4f46\u5176\u5b9e\u91c7\u7528\u6210\u719f\u7684\u65b9\u6cd5\u5df2\u7ecf\u8db3\u591f\u4e86\uff0c\u6240\u4ee5\u4e00\u822c\u6765\u8bf4\u673a\u5668\u5b66\u4e60\u5bf9\u4e8e\u5de5\u7a0b\u4eba\u5458\u7684\u4e3b\u8981\u95ee\u9898\u5c31\u5728\u4e8e\u5982\u4f55\u8fdb\u884c\u6570\u636e\u5904\u7406\u3002
    \n\u6570\u636e\u91cf\u8f83\u5927\u65f6\u9700\u8981\u8bf8\u5982GPU\u3001TPU\u7b49\u8ba1\u7b97\u673a\u8d44\u6e90\u8fdb\u884c\u8ba1\u7b97\uff0c\u4e5f\u53ef\u4ee5\u8fdb\u884c\u5206\u5e03\u5f0f\uff0c\u4e0d\u8fc7\u8fd8\u662f\u90a3\u53e5\u8bdd\uff0c\u5927\u591a\u6570\u573a\u666f\u57fa\u672c\u4e0a\u4e0d\u9700\u8981\u8003\u8651\u90a3\u4e48\u591a\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"
    https:\/\/blog.csdn.net\/weixin_45638544\/article\/details\/1… <\/p>\n","protected":false},"author":28,"featured_media":3601,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[7],"tags":[],"class_list":["post-3543","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-7"],"_links":{"self":[{"href":"https:\/\/aiwellbore.com\/index.php?rest_route=\/wp\/v2\/posts\/3543","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aiwellbore.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aiwellbore.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aiwellbore.com\/index.php?rest_route=\/wp\/v2\/users\/28"}],"replies":[{"embeddable":true,"href":"https:\/\/aiwellbore.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=3543"}],"version-history":[{"count":12,"href":"https:\/\/aiwellbore.com\/index.php?rest_route=\/wp\/v2\/posts\/3543\/revisions"}],"predecessor-version":[{"id":3566,"href":"https:\/\/aiwellbore.com\/index.php?rest_route=\/wp\/v2\/posts\/3543\/revisions\/3566"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/aiwellbore.com\/index.php?rest_route=\/wp\/v2\/media\/3601"}],"wp:attachment":[{"href":"https:\/\/aiwellbore.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3543"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aiwellbore.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3543"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aiwellbore.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3543"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}