{"id":285,"date":"2017-01-23T16:50:00","date_gmt":"2017-01-23T08:50:00","guid":{"rendered":"http:\/\/www.mrtblog.cn\/?p=285"},"modified":"2023-03-05T14:40:58","modified_gmt":"2023-03-05T06:40:58","slug":"%e5%86%b3%e7%ad%96%e6%a0%91%e7%ae%97%e6%b3%95%e5%8f%8a%e5%ae%9e%e8%b7%b5","status":"publish","type":"post","link":"http:\/\/www.mrtblog.cn\/?p=285","title":{"rendered":"\u51b3\u7b56\u6811\u7b97\u6cd5\u53ca\u5b9e\u8df5"},"content":{"rendered":"<div class='epvc-post-count'><span class='epvc-eye'><\/span>  <span class=\"epvc-count\"> 1,374<\/span><span class='epvc-label'> Views<\/span><\/div><p><em>\u53c2\u8003\u4e66\u7c4d\u300a\u673a\u5668\u5b66\u4e60\u5b9e\u6218\u300b<\/em><\/p>\n<h3>\u51b3\u7b56\u6811\u7b97\u6cd5\u6982\u8ff0<\/h3>\n<p>K-\u8fd1\u90bb\u7b97\u6cd5\u53ef\u4ee5\u5b8c\u6210\u5f88\u591a\u5206\u7c7b\u4efb\u52a1\uff0c\u4f46\u662f\u5b83\u6700\u5927\u7684\u7f3a\u70b9\u5c31\u662f\u65e0\u6cd5\u7ed9\u51fa\u6570\u636e\u7684\u5185\u5728\u542b\u4e49\uff0c\u51b3\u7b56\u6811\u7684\u4e3b\u8981\u4f18\u52bf\u5c31\u5728\u4e8e\u6570\u636e\u5f62\u5f0f\u975e\u5e38\u5bb9\u6613\u7406\u89e3\u3002<\/p>\n<p>\u51b3\u7b56\u6811\u7b97\u6cd5\u80fd\u591f\u8bfb\u53d6\u6570\u636e\u96c6\u5408\uff0c\u6784\u5efa\u7c7b\u4f3c\u4e8e\u56fe3-1\u7684\u51b3\u7b56\u6811\u3002\u51b3\u7b56\u6811\u5f88\u591a\u4efb\u52a1\u90fd\u662f\u4e3a\u4e86\u6570\u636e\u4e2d\u6240\u8574\u542b\u7684\u77e5\u8bc6\u4fe1\u606f\uff0c\u56e0\u6b64\u51b3\u7b56\u6811\u53ef\u4ee5\u4f7f\u7528\u4e0d\u719f\u6089\u7684\u6570\u636e\u96c6\u5408\uff0c\u5e76\u4ece\u4e2d\u63d0\u53d6\u51fa\u4e00\u7cfb\u5217\u89c4\u5219\uff0c\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\u6700\u7ec8\u5c06\u4f7f\u7528\u8fd9\u4e9b\u673a\u5668\u4ece\u6570\u636e\u96c6\u4e2d\u521b\u9020\u7684\u89c4\u5219\u3002\u4e13\u5bb6\u7cfb\u7edf\u4e2d\u7ecf\u5e38\u4f7f\u7528\u51b3\u7b56\u6811\uff0c\u800c\u4e14\u51b3\u7b56\u6811\u7ed9\u51fa\u7ed3\u679c\u5f80\u5f80\u53ef\u4ee5\u5339\u654c\u5728\u5f53\u524d\u9886\u57df\u5177\u6709\u51e0\u5341\u5e74\u5de5\u4f5c\u7ecf\u9a8c\u7684\u4eba\u7c7b\u4e13\u5bb6\u3002<a href=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-p3-1.png\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-289\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-p3-1.png\" alt=\"\" width=\"632\" height=\"550\" srcset=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-p3-1.png 632w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-p3-1-300x261.png 300w\" sizes=\"auto, (max-width: 632px) 100vw, 632px\" \/><\/a><\/p>\n<p><strong>\u51b3\u7b56\u6811\u7684\u4e00\u822c\u6d41\u7a0b<\/strong><br \/>\n(1)\u6536\u96c6\u6570\u636e\uff1a\u53ef\u4ee5\u4f7f\u7528\u4efb\u4f55\u65b9\u6cd5\u3002<br \/>\n(2)\u51c6\u5907\u6570\u636e\uff1a\u6811\u6784\u9020\u7b97\u6cd5\u53ea\u9002\u7528\u4e8e\u6807\u79f0\u578b\u6570\u636e\uff0c\u56e0\u6b64\u6570\u503c\u578b\u6570\u636e\u5fc5\u987b\u79bb\u6563\u5316\u3002<br \/>\n(3)\u5206\u6790\u6570\u636e\uff1a\u53ef\u4ee5\u4f7f\u7528\u4efb\u4f55\u65b9\u6cd5\uff0c\u6784\u9020\u6811\u5b8c\u6210\u4e4b\u540e\uff0c\u6211\u4eec\u5e94\u8be5\u68c0\u67e5\u56fe\u5f62\u662f\u5426\u7b26\u5408\u9884\u671f\u3002<br \/>\n(4)\u8bad\u7ec3\u7b97\u6cd5\uff1a\u6784\u9020\u6811\u7684\u6570\u636e\u7ed3\u6784\u3002<br \/>\n(5)\u6d4b\u8bd5\u7b97\u6cd5\uff1a\u4f7f\u7528\u7ecf\u9a8c\u6811\u8ba1\u7b97\u9519\u8bef\u7387\u3002<br \/>\n(6)\u4f7f\u7528\u7b97\u6cd5\uff1a\u6b64\u6b65\u9aa4\u53ef\u4ee5\u9002\u7528\u4e8e\u4efb\u4f55\u76d1\u7763\u5b66\u4e60\u7b97\u6cd5\uff0c\u800c\u4f7f\u7528\u51b3\u7b56\u6811\u53ef\u4ee5\u66f4\u597d\u5730\u7406\u89e3\u6570\u636e \u7684\u5185\u5728\u542b\u4e49\u3002<\/p>\n<p>\u4e00\u4e9b\u51b3\u7b56\u6811\u7b97\u6cd5\u91c7\u7528\u4e8c\u5206\u6cd5\u5212\u5206\u6570\u636e\uff0c\u672c\u4e66\u5e76\u4e0d\u91c7\u7528\u8fd9\u79cd\u65b9\u6cd5\u3002\u5982\u679c\u4f9d\u636e\u67d0\u4e2a\u5c5e\u6027\u5212\u5206\u6570\u636e\u5c06\u4f1a\u4ea7\u751f4\u4e2a\u53ef\u80fd\u7684\u503c\uff0c\u6211\u4eec\u5c06\u628a\u6570\u636e\u5212\u5206\u6210\u56db\u5757\uff0c\u5e76\u521b\u5efa\u56db\u4e2a\u4e0d\u540c\u7684\u5206\u652f\u3002\u6bcf\u6b21\u5212\u5206\u6570\u636e\u96c6\u65f6\u6211\u4eec\u53ea\u9009\u53d6\u4e00\u4e2a\u7279\u5f81\u5c5e\u6027\uff0c\u5982\u679c\u8bad\u7ec3\u96c6\u4e2d\u5b58\u572820\u4e2a\u7279\u5f81\uff0c\u7b2c\u4e00\u6b21\u6211\u4eec\u9009\u62e9\u54ea\u4e2a\u7279\u5f81\u4f5c\u4e3a\u5212\u5206\u7684\u53c2\u8003\u5c5e\u6027\u5462\uff1f<\/p>\n<p>\u88683-1\u7684\u6570\u636e\u5305\u542b5\u4e2a\u6d77\u6d0b\u52a8\u7269\uff0c\u7279\u5f81\u5305\u62ec\uff1a\u4e0d\u6d6e\u51fa\u6c34\u9762\u662f\u5426\u53ef\u4ee5\u751f\u5b58\uff0c\u4ee5\u53ca\u662f\u5426\u6709\u811a\u8e7c\u3002\u6211\u4eec\u53ef\u4ee5\u5c06\u8fd9\u4e9b\u52a8\u7269\u5206\u6210\u4e24\u7c7b\uff1a\u9c7c\u7c7b\u548c\u975e\u9c7c\u7c7b\u3002\u73b0\u5728\u6211\u4eec\u60f3\u8981\u51b3\u5b9a\u4f9d\u636e\u7b2c\u4e00\u4e2a\u7279\u5f81\u8fd8\u662f\u7b2c\u4e8c\u4e2a\u7279\u5f81\u5212\u5206\u6570\u636e\u3002\u5728\u56de\u7b54\u8fd9\u4e2a\u95ee\u9898\u4e4b\u524d\uff0c\u6211\u4eec\u5fc5\u987b\u91c7\u7528\u91cf\u5316\u7684\u65b9\u6cd5\u5224\u65ad\u5982\u4f55\u5212\u5206\u6570\u636e\u3002<a href=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-t3-1.png\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-290\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-t3-1.png\" alt=\"\" width=\"1122\" height=\"300\" srcset=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-t3-1.png 1122w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-t3-1-300x80.png 300w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-t3-1-768x205.png 768w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-t3-1-1024x274.png 1024w\" sizes=\"auto, (max-width: 1122px) 100vw, 1122px\" \/><\/a><\/p>\n<p><strong>\u4fe1\u606f\u5897\u76ca<\/strong><br \/>\n\u5212\u5206\u6570\u636e\u96c6\u7684\u5927\u539f\u5219\u662f\uff1a\u5c06\u65e0\u5e8f\u7684\u6570\u636e\u53d8\u5f97\u66f4\u52a0\u6709\u5e8f\u3002\u6211\u4eec\u53ef\u4ee5\u4f7f\u7528\u591a\u79cd\u65b9\u6cd5\u5212\u5206\u6570\u636e\u96c6\uff0c\u4f46\u662f\u6bcf\u79cd\u65b9\u6cd5\u90fd\u6709\u5404\u81ea\u7684\u4f18\u7f3a\u70b9\u3002\u7ec4\u7ec7\u6742\u4e71\u65e0\u7ae0\u6570\u636e\u7684\u4e00\u79cd\u65b9\u6cd5\u5c31\u662f\u4f7f\u7528\u4fe1\u606f\u8bba\u5ea6\u91cf\u4fe1\u606f\uff0c\u4fe1\u606f\u8bba\u662f\u91cf\u5316\u5904\u7406\u4fe1\u606f\u7684\u5206\u652f\u79d1\u5b66\u3002\u6211\u4eec\u53ef\u4ee5\u5728\u5212\u5206\u6570\u636e\u4e4b\u524d\u4f7f\u7528\u4fe1\u606f\u8bba\u91cf\u5316\u5ea6\u91cf\u4fe1\u606f\u7684\u5185\u5bb9\u3002<br \/>\n\u5728\u5212\u5206\u6570\u636e\u96c6\u4e4b\u524d\u4e4b\u540e\u4fe1\u606f\u53d1\u751f\u7684\u53d8\u5316\u79f0\u4e3a\u4fe1\u606f\u589e\u76ca\uff0c\u77e5\u9053\u5982\u4f55\u8ba1\u7b97\u4fe1\u606f\u589e\u76ca\uff0c\u6211\u4eec\u5c31\u53ef\u4ee5\u8ba1\u7b97\u6bcf\u4e2a\u7279\u5f81\u503c\u5212\u5206\u6570\u636e\u96c6\u83b7\u5f97\u7684\u4fe1\u606f\u589e\u76ca\uff0c\u83b7\u5f97\u4fe1\u606f\u589e\u76ca\u6700\u9ad8\u7684\u7279\u5f81\u5c31\u662f\u6700\u597d\u7684\u9009\u62e9\u3002<\/p>\n<p>\u71b5\u5b9a\u4e49\u4e3a\u4fe1\u606f\u7684\u671f\u671b\u503c\uff0c\u5728\u660e\u6670\u8fd9\u4e2a\u6982\u5ff5\u4e4b\u524d\uff0c\u6211\u4eec\u5fc5\u987b\u77e5\u9053\u4fe1\u606f\u7684\u5b9a\u4e49\u3002\u5982\u679c\u5f85\u5206\u7c7b\u7684\u4e8b\u52a1\u53ef\u80fd\u5212\u5206\u5728\u591a\u4e2a\u5206\u7c7b\u4e4b\u4e2d\uff0c\u5219\u7b26\u53f7xi\u7684\u4fe1\u606f\u5b9a\u4e49\u4e3a\uff1a<img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-293\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-formula.png\" alt=\"\" width=\"209\" height=\"51\">\u5176\u4e2dp(xi)\u662f\u9009\u62e9\u8be5\u5206\u7c7b\u7684\u6982\u7387\u3002<\/p>\n<p>\u4e3a\u4e86\u8ba1\u7b97\u71b5\uff0c\u6211\u4eec\u9700\u8981\u8ba1\u7b97\u6240\u6709\u7c7b\u522b\u6240\u6709\u53ef\u80fd\u503c\u5305\u542b\u7684\u4fe1\u606f\u671f\u671b\u503c\uff0c\u901a\u8fc7\u4e0b\u9762\u7684\u516c\u5f0f\u5f97\u5230\uff1a<a href=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-formula-1.png\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-296\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-formula-1.png\" alt=\"\" width=\"295\" height=\"46\"><\/a><\/p>\n<h4>\u4f7f\u7528Python\u5b9e\u73b0\u51b3\u7b56\u6811\u7b97\u6cd5<\/h4>\n<p>\u521b\u5efa\u6587\u4ef6Tree.py<\/p>\n<pre class=\"highlight\"><code class=\"language-python line-numbers\">\n#coding:utf-8\nfrom math import log\ndef createDataSet():#\u521b\u5efa\u6570\u636e\u96c6\n    dataSet = [#\u4e0d\u6d6e\u51fa\u6c34\u9762\u662f\u5426\u53ef\u4ee5\u751f\u5b58\uff0c\u662f\u5426\u6709\u811a\u8e1d\uff0c\u5c5e\u4e8e\u9c7c\u7c7b\n        [1,1,&#039;yes&#039;],\n        [1,1,&#039;yes&#039;],\n        [1,0,&#039;no&#039;],\n        [0,1,&#039;no&#039;],\n        [0,1,&#039;no&#039;]\n    ]\n    labels = [&#039;no surfacing&#039;,&#039;flippers&#039;]\n    return dataSet,labels\ndef calcShannonEnt(dataSet):#\u8ba1\u7b97\u71b5\n    numEntries=len(dataSet)#\u6570\u636e\u96c6\u603b\u6570   #\u6570\u636e\u96c6\u683c\u5f0f\u4e3a{\u6761\u4ef61\uff0c\u6761\u4ef62...\uff0c\u5206\u7c7b\u7ed3\u679c}\n    labelCounts = {}\n    for featVec in dataSet:\n        currentLabel = featVec[-1]#\u5f97\u5230\u5206\u7c7b\u7ed3\u679c\n        if currentLabel not in labelCounts.keys():#\u65b0\u7684\u5206\u7c7b\n            labelCounts[currentLabel] = 0#\u8bb0\u5f55\u5206\u7c7b\n            labelCounts[currentLabel] += 1#\u8be5\u5206\u7c7b\u603b\u6570+1\n    shannonEnt = 0.0\n    for key in labelCounts:\n        prob = float(labelCounts[key])\/numEntries#\u4fe1\u606f\u503c\n        shannonEnt -= prob * log(prob,2)#\u71b5\n    return shannonEnt\n<\/code><\/pre>\n<p>\u6d4b\u8bd5\u8f93\u51fa\u6570\u636e\u96c6\u53ca\u8f93\u51fa\u71b5\u7684\u8ba1\u7b97\u7ed3\u679c<br \/>\n<a href=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-example.png\"><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-example.png\" alt=\"\" width=\"818\" height=\"286\" class=\"alignnone size-full wp-image-297\" srcset=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-example.png 818w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-example-300x105.png 300w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-example-768x269.png 768w\" sizes=\"auto, (max-width: 818px) 100vw, 818px\" \/><\/a><br \/>\n\u5f97\u5230\u71b5\u4e4b\u540e\uff0c\u6211\u4eec\u5c31\u53ef\u4ee5\u6309\u7167\u83b7\u53d6\u6700\u5927\u4fe1\u606f\u589e\u76ca\u7684\u65b9\u6cd5\u5212\u5206\u6570\u636e\u96c6<\/p>\n<p><strong>\u5212\u5206\u6570\u636e\u96c6<\/strong><br \/>\n\u4e0a\u8282\u6211\u4eec\u5b66\u4e60\u4e86\u5982\u4f55\u5ea6\u91cf\u6570\u636e\u96c6\u7684\u65e0\u5e8f\u7a0b\u5ea6\uff0c\u5206\u7c7b\u7b97\u6cd5\u9664\u4e86\u9700\u8981\u6d4b\u91cf\u4fe1\u606f\u71b5\uff0c\u8fd8\u9700\u8981\u5212\u5206\u6570\u636e\u96c6\uff0c\u5ea6\u91cf\u82b1\u8d39\u6570\u636e\u96c6\u7684\u71b5\uff0c\u4ee5\u4fbf\u5224\u65ad\u5f53\u524d\u662f\u5426\u6b63\u786e\u5730\u5212\u5206\u4e86\u6570\u636e\u96c6\u3002\u6211\u4eec\u5c06\u5bf9\u6bcf\u4e2a\u7279\u5f81\u5212\u5206\u6570\u636e\u96c6\u7684\u7ed3\u679c\u8ba1\u7b97\u4e00\u6b21\u4fe1\u606f\u71b5\uff0c\u7136\u540e\u5224\u65ad\u6309\u7167\u54ea\u4e2a\u7279\u5f81\u5212\u5206\u6570\u636e\u96c6\u662f\u6700\u597d\u7684\u5212\u5206\u65b9\u5f0f\u3002\u60f3\u8c61\u4e00\u4e2a\u5206\u5e03\u5728\u4e8c\u7ef4\u7a7a\u95f4\u7684\u6570\u636e\u6563\u70b9\u56fe\uff0c\u9700\u8981\u5728\u6570\u636e\u4e4b\u95f4\u5212\u6761\u7ebf\uff0c\u5c06\u5b83\u4eec\u5206\u6210\u4e24\u90e8\u5206\uff0c\u6211\u4eec\u5e94\u8be5\u6309\u7167x\u8f74\u8fd8\u662fy\u8f74\u5212\u7ebf\u5462\uff1f\u7b54\u6848\u5c31\u662f\u672c\u8282\u8bb2\u8ff0\u7684\u5185\u5bb9\u3002\u8981\u5212\u5206\u6570\u636e\u96c6\uff0c\u6253\u5f00\u6587\u672c\u7f16\u8f91\u5668\uff0c\u5728Tree.py\u6587\u4ef6\u4e2d\u6dfb\u52a0\u4e0b\u5217\u7684\u4ee3\u7801\uff1a<\/p>\n<pre class=\"highlight\"><code class=\"language-python line-numbers\">\ndef splitDataSet(dataSet,axis,value):#\u5212\u5206\u6570\u636e\u96c6\n    retDataSet = []#\u751f\u6210\u65b0\u7684list\n    for featVec in dataSet:\n        if featVec[axis] == value:#\u6ee1\u8db3\u6761\u4ef6\u5219\u628a\u8be5\u6761\u6570\u636e\u63d0\u53d6\u51fa\u6765\n            reducedFeatVec = featVec[:axis]\n            reducedFeatVec.extend(featVec[axis+1:])\n            retDataSet.append(reducedFeatVec)\n    return retDataSet#\u8fd4\u56de\u65b0\u7684\u6570\u636e\u96c6\n<\/code><\/pre>\n<p>\u6309\u7167\u4e0d\u540c\u6761\u4ef6\u5212\u5206\u6d4b\u8bd5\uff1a<br \/>\n<a href=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-example-1.png\"><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-example-1.png\" alt=\"\" width=\"802\" height=\"304\" class=\"alignnone size-full wp-image-299\" srcset=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-example-1.png 802w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-example-1-300x114.png 300w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/01\/Decision-Tree-example-1-768x291.png 768w\" sizes=\"auto, (max-width: 802px) 100vw, 802px\" \/><\/a><\/p>\n<p>\u63a5\u4e0b\u6765\u6211\u4eec\u5c06\u904d\u5386\u6574\u4e2a\u6570\u636e\u96c6\uff0c\u5faa\u73af\u8ba1\u7b97\u9999\u519c\u71b5\u548csplitDataSet()\u51fd\u6570\uff0c\u627e\u5230\u6700\u597d\u7684\u7279\u5f81\u5212\u5206\u65b9\u5f0f\u3002\u71b5\u8ba1\u7b97\u5c06\u4f1a\u544a\u8bc9\u6211\u4eec\u5982\u4f55\u5212\u5206\u6570\u636e\u96c6\u662f\u6700\u597d\u7684\u6570\u636e\u7ec4\u7ec7\u65b9\u5f0f\u3002<\/p>\n<hr>\n<p><strong>\u9009\u62e9\u6700\u597d\u7684\u6570\u636e\u96c6\u5212\u5206\u65b9\u5f0f<\/strong><\/p>\n<p>\u5728Tree.py\u4e2d\u6dfb\u52a0\u5982\u4e0b\u4ee3\u7801\uff1a<\/p>\n<pre class=\"highlight\"><code class=\"language-python line-numbers\">\ndef chooseBestFeatureToSplit(dataSet):\n    numFeatures = len(dataSet[0]) - 1#\u5f97\u5230\u7279\u5f81\u6570(\u4e0d\u5305\u62ec\u5206\u7c7b)\n    baseEntropy = calcShannonEnt(dataSet)#\u8ba1\u7b97\u4fe1\u606f\u71b5\n    bestInfoGain = 0.0\n    bestFeature = -1\n    for i in range(numFeatures):\n        featList = [example[i] for example in dataSet]#\u5f97\u5230\u6570\u636e\u96c6\u6bcf\u4e2a\u6570\u636e\u7684\u7b2ci\u4e2a\u7279\u5f81\u503c\n        uniqueVals = set(featList)#\u5f97\u5230\u7b2ci\u4e2a\u7279\u5f81\u7684\u6240\u6709\u975e\u91cd\u590d\u503c\n        newEntropy = 0.0\n        for value in uniqueVals:#\u8ba1\u7b97\u6309\u7167\u7b2ci\u4e2a\u7279\u5f81\u6309\u7167\u6240\u6709\u975e\u91cd\u590d\u503c\u5212\u5206\u65f6\u5305\u542b\u7684\u4fe1\u606f\u71b5\n            subDataSet = splitDataSet(dataSet,i,value)\n            prob = len(subDataSet) \/ float(len(dataSet))\n            newEntropy += prob * calcShannonEnt(subDataSet)\n        infoGain = baseEntropy - newEntropy\n        if (infoGain &gt; bestInfoGain):\n            bestInfoGain = infoGain\n            bestFeature = i\n    return bestFeature\n<\/code><\/pre>\n<p>chooseBestFeatureToSplit\u51fd\u6570\u8fd4\u56de\u7ed3\u679c\u544a\u8bc9\u6211\u4eec\u5e94\u8be5\u6309\u7167\u54ea\u4e00\u4e2a\u7279\u5f81\u5212\u5206\u5305\u542b\u7684\u4fe1\u606f\u6700\u5927\uff0c\u7ed3\u679c\u4e3a0\uff08\u7b2c\u4e00\u4e2a\u7279\u5f81\uff09\uff1a<br \/>\n<a href=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-2.png\"><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-2.png\" alt=\"\" width=\"1189\" height=\"291\" class=\"alignnone size-full wp-image-330\" srcset=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-2.png 1189w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-2-300x73.png 300w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-2-768x188.png 768w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-2-1024x251.png 1024w\" sizes=\"auto, (max-width: 1189px) 100vw, 1189px\" \/><\/a><\/p>\n<p>\u90a3\u4e48\u8fd9\u4e2a\u5212\u5206\u7ed3\u679c\u662f\u5426\u6709\u610f\u4e49\u5462\uff1f\u5982\u679c\u6211\u4eec\u6309\u7167\u7b2c\u4e00\u4e2a\u7279\u5f81\u5212\u5206\u6570\u636e\u96c6\uff0c\u7b2c\u4e00\u4e2a\u7279\u5f81\u4e3a1\u7684\u6d77\u6d0b\u751f\u7269\u6709\u4e24\u4e2a\u9c7c\u7c7b\uff0c\u4e00\u4e2a\u975e\u9c7c\u7c7b\uff1b\u53e6\u4e00\u4e2a\u5206\u7ec4\u5219\u5168\u90e8\u5c5e\u4e8e\u975e\u9c7c\u7c7b\u3002\u5982\u679c\u6309\u7167\u7b2c\u4e8c\u4e2a\u7279\u5f81\u5212\u5206\uff0c\u5219\u7b2c\u4e00\u4e2a\u5206\u7ec4\u6709\u4e24\u4e2a\u9c7c\u7c7b\uff0c\u4e24\u4e2a\u975e\u9c7c\u7c7b\uff1b\u53e6\u4e00\u4e2a\u5206\u7ec4\u6709\u4e00\u4e2a\u975e\u9c7c\u7c7b\u3002\u663e\u7136\u6309\u7167\u7b2c\u4e00\u4e2a\u7279\u5f81\u5212\u5206\u66f4\u597d\u3002<\/p>\n<p><strong>\u9012\u5f52\u6784\u5efa\u51b3\u7b56\u6811<\/strong><\/p>\n<p>\u4e4b\u524d\u7684\u6b65\u9aa4\u6211\u4eec\u5df2\u7ecf\u80fd\u591f\u6309\u7167\u6700\u597d\u7684\u5c5e\u6027\u5212\u5206\u6570\u636e\u96c6\uff0c\u7531\u4e8e\u7279\u5f81\u53ef\u80fd\u591a\u4e8e\u4e24\u4e2a\uff0c\u9009\u62e9\u4e00\u4e2a\u7279\u5f81\u540e\u5e94\u8be5\u4ece\u5269\u4e0b\u7684\u7279\u5f81\u7ee7\u7eed\u9009\u62e9\u6700\u597d\u7684\u5212\u5206\u7279\u5f81\uff0c\u76f4\u5230\u6240\u6709\u7279\u5f81\u90fd\u88ab\u8003\u8651\u5728\u5185\uff0c\u56e0\u6b64\u6211\u4eec\u53ef\u4ee5\u91c7\u7528\u9012\u5f52\u7684\u539f\u5219\u5904\u7406\u6570\u636e\u96c6\u3002\u53c2\u89c1\u56fe3-2<br \/>\n<a href=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-p3-2.png\"><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-p3-2.png\" alt=\"\" width=\"622\" height=\"535\" class=\"alignnone size-full wp-image-332\" srcset=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-p3-2.png 622w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-p3-2-300x258.png 300w\" sizes=\"auto, (max-width: 622px) 100vw, 622px\" \/><\/a><\/p>\n<p>\u5982\u679c\u6570\u636e\u96c6\u5df2\u7ecf\u5904\u7406\u4e86\u6240\u6709\u7684\u5c5e\u6027\uff0c\u4f46\u662f\u7c7b\u6807\u7b7e\u4f9d\u7136\u4e0d\u662f\u552f\u4e00\u7684\uff0c\u6b64\u65f6\u6211\u4eec\u91c7\u7528\u591a\u6570\u8868\u51b3\u7684\u65b9\u6cd5\u51b3\u5b9a\u5176\u6807\u7b7e\u3002<br \/>\n\u5728Tree.py\u4e2d\u6dfb\u52a0\u5982\u4e0b\u4ee3\u7801\uff1a<\/p>\n<pre class=\"highlight\"><code class=\"language-python line-numbers\">\nimport operator\ndef majorityCnt(classList):\n    classCount = {}\n    for vote in classList:\n        if vote not in classCount.keys():\n            classCount[vote]=0\n        classCount[vote]+=1\n    sortedClassCount = sorted(classCount.iteritems(),key=operator.itemgetter(1),reverse=True)\n    return sortedClassCount[0][0]\n<\/code><\/pre>\n<p>\u4e3a\u4e86\u521b\u5efa\u6811\uff0c\u5728Tree.py\u4e2d\u6dfb\u52a0\u5982\u4e0b\u4ee3\u7801\uff1a<\/p>\n<pre class=\"highlight\"><code class=\"language-python line-numbers\">\ndef createTree(dataSet,labels):\n    classList = [example[-1] for example in dataSet]#\u5f97\u5230label\u6807\u7b7e\n    if classList.count(classList[0])==len(classList):#\u6240\u6709\u6807\u7b7e\u90fd\u662f\u4e00\u6837\u7684\uff0c\u76f4\u63a5\u8fd4\u56de\n        return classList[0]\n    if len(dataSet[0]) == 1:#\u6240\u6709\u7279\u5f81\u904d\u53861\u6b21\n        return majorityCnt(classList)\n    bestFeat = chooseBestFeatureToSplit(dataSet)#\u6700\u597d\u5212\u5206\u7279\u5f81\n    bestFeatLabel = labels[bestFeat]#\u8be5\u7279\u5f81\u5bf9\u5e94\u7684\u6807\u7b7e\n    myTree = {bestFeatLabel:{}}\n    del(labels[bestFeat])#\u4ecelabels\u4e2d\u5220\u9664\u8be5\u7279\u5f81\n    featValues = [example[bestFeat] for example in dataSet]#\u5728dataSet\u4e2d\u5f97\u5230\u8be5\u7279\u5f81\u7684\u6240\u6709\u503c\n    uniqueVals = set(featValues)#\u5f97\u5230\u8be5\u7279\u5f81\u7684\u4e0d\u91cd\u590d\u503c\n    for value in uniqueVals:\n        subLabels = labels[:]#\u9664\u53bb\u8be5\u7279\u5f81\u540e\u8fd8\u5269\u4f59\u7684\u5176\u4ed6\u7279\u5f81\n        #\u5c06\u5269\u4f59\u7279\u5f81\u5199\u4f5c\u5f53\u524d\u7279\u5f81\u7684\u5b50\u6811\n        myTree[bestFeatLabel][value] = createTree(splitDataSet(dataSet,bestFeat,value),subLabels)\n    return myTree\n<\/code><\/pre>\n<p>\u4ee5\u9012\u5f52\u7684\u65b9\u5f0f\u521b\u5efa\uff0c\u6bcf\u4e00\u6b21\u5224\u65ad\u4e00\u4e2a\u7279\u5f81\uff0c\u5c06\u5269\u4f59\u7279\u5f81\u505a\u4e3a\u8be5\u7279\u5f81\u7684\u5206\u652f\uff0c\u76f4\u5230\u6240\u6709\u7279\u5f81\u904d\u5386\u5b8c\uff1b\u82e5\u6709\u4e0d\u80fd\u51b3\u65ad\u7684\u6807\u7b7e\u5219\u6309\u591a\u6570\u8868\u51b3\u51b3\u5b9a\u3002<br \/>\n\u51b3\u7b56\u6811\u5efa\u7acb\u5b8c\u6210\u540e\u8f93\u51fa\u6d4b\u8bd5\uff1a<br \/>\n<a href=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-3.png\"><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-3.png\" alt=\"\" width=\"1180\" height=\"381\" class=\"alignnone size-full wp-image-335\" srcset=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-3.png 1180w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-3-300x97.png 300w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-3-768x248.png 768w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-3-1024x331.png 1024w\" sizes=\"auto, (max-width: 1180px) 100vw, 1180px\" \/><\/a><\/p>\n<p><strong>\u6811\u7684\u56fe\u5f62\u5316\u663e\u793a\uff1a\u7565<\/strong><\/p>\n<p><strong>\u6d4b\u8bd5\u548c\u5b58\u50a8\u5206\u7c7b\u5668<\/strong><\/p>\n<p>\u4e4b\u524d\u4e3b\u8981\u5185\u5bb9\u4e3a\u5982\u4f55\u521b\u5efa\u51b3\u7b56\u6811\uff0c\u672c\u8282\u5c06\u4f7f\u7528\u51b3\u7b56\u6811\u6784\u5efa\u5206\u7c7b\u5668\u3002<br \/>\n\u5728Tree.py\u4e2d\u6dfb\u52a0\u5206\u7c7b\u4ee3\u7801\uff1a<\/p>\n<pre class=\"highlight\"><code class=\"language-python line-numbers\">\ndef classify(inputTree,featLabels,testVec):\n    firstStr = inputTree.keys()[0]#\u5f97\u5230\u7b2c\u4e00\u4e2a\u7279\u5f81\u7684\u540d\u79f0\n    secondDict = inputTree[firstStr]#\u5f97\u5230\u4e0e\u7b2c\u4e00\u4e2a\u7279\u5f81\u5bf9\u5e94\u7684\u5b57\u5178\n    featIndex = featLabels.index(firstStr)#\u5f97\u5230\u7b2c\u4e00\u4e2a\u7279\u5f81\u5728\u7279\u5f81\u6807\u7b7e\u4e2d\u7684\u5e8f\u53f7\n    for key in secondDict.keys():#\u904d\u5386\u8be5\u7279\u5f81\u7684\u5bf9\u5e94\u5b57\u5178\n        if testVec[featIndex] == key:#\u8be5\u8f93\u5165\u6570\u636e\u5bf9\u5e94\u7279\u5f81\u503c\u7b26\u5408\n            if type(secondDict[key]).__name__ == &#039;dict&#039;:#\u5b50\u7ea7\u4f9d\u7136\u662f\u5b57\u5178\uff0c\u7ee7\u7eed\u9012\u5f52\n                classLabel = classify(secondDict[key],featLabels,testVec)\n            else:#\u5b50\u7ea7\u4e3a\u53f6\u5b50\uff0c\u53ef\u4ee5\u76f4\u63a5\u5f97\u51fa\u7ed3\u8bba\n                classLabel = secondDict[key]\n    return classLabel\n<\/code><\/pre>\n<p>\u867d\u7136\u6811\u4e2d\u5212\u5206\u597d\u4e86\u7279\u5f81\u5206\u7c7b\u7684\u7ed3\u6784\uff0c\u4f46\u662f\u8ba1\u7b97\u673a\u5e76\u4e0d\u77e5\u9053\u7279\u5f81\u5728\u7279\u5f81\u5206\u7ec4\u4e2d\u7684\u4f4d\u7f6e\uff0c\u4f8b\u5982&#8217;no surfacing&#8217;\u5728\u6d4b\u8bd5\u6570\u636e\u4e2d\u5230\u5e95\u662f\u5728\u7b2c\u4e00\u4e2a\u4f4d\u7f6e\u8fd8\u662f\u7b2c2\u4e2a\u4f4d\u7f6e\uff1f\u6240\u4ee5\u5206\u7c7b\u5668\u7684\u7b2c\u4e00\u6b65\u662f\u5224\u65ad\u51fa\u5404\u4e2a\u7279\u5f81\u5728\u6e90\u6570\u636e\u4e2d\u7684\u4f4d\u7f6e\uff0c\u4e4b\u540e\u518d\u904d\u5386\u6574\u68f5\u6811\uff0c\u627e\u5230\u6d4b\u8bd5\u6570\u636e\u7684\u5bf9\u5e94\u6807\u7b7e\u3002<br \/>\n<a href=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-4.png\"><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-4.png\" alt=\"\" width=\"1202\" height=\"565\" class=\"alignnone size-full wp-image-338\" srcset=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-4.png 1202w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-4-300x141.png 300w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-4-768x361.png 768w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-4-1024x481.png 1024w\" sizes=\"auto, (max-width: 1202px) 100vw, 1202px\" \/><\/a><br \/>\n\u53ef\u4ee5\u770b\u5230\u5bf9\u8f93\u5165\u7684\u6570\u636e\uff0c\u51b3\u7b56\u6811\u80fd\u505a\u51fa\u6b63\u786e\u7684\u5224\u65ad\uff0c\u5373\u4f7f\u662f\u8bad\u7ec3\u6570\u636e\u6ca1\u6709\u7684\u7ec4\u5408\uff0c\u4e5f\u80fd\u4f5c\u51fa\u76f8\u5e94\u5224\u65ad\u3002<\/p>\n<p><strong>\u51b3\u7b56\u6811\u7684\u5b58\u50a8<\/strong><\/p>\n<p>\u6784\u5efa\u51b3\u7b56\u6811\u662f\u4e00\u4e2a\u5f88\u8017\u65f6\u7684\u4efb\u52a1\uff0c\u5982\u679c\u6570\u636e\u91cf\u8f83\u5927\uff0c\u4f1a\u82b1\u8d39\u66f4\u591a\u65f6\u95f4\u3002\u800c\u4e0d\u540c\u4e8eKNN\u7b97\u6cd5\uff0c\u51b3\u7b56\u6811\u7684\u597d\u5904\u662f\u80fd\u591f\u4f7f\u7528\u521b\u5efa\u597d\u7684\u51b3\u7b56\u6811\u89e3\u51b3\u5206\u7c7b\u95ee\u9898\uff0c\u6211\u4eec\u53ef\u4ee5\u5728\u521b\u5efa\u597d\u6811\u540e\u4fdd\u5b58\u5728\u672c\u5730\u78c1\u76d8\uff0c\u65b9\u4fbf\u4e0b\u6b21\u4f7f\u7528\u3002<br \/>\n\u5728Tree.py\u4e2d\u6dfb\u52a0\u5982\u4e0b\u4ee3\u7801\uff1a<\/p>\n<pre class=\"highlight\"><code class=\"language-python line-numbers\">\ndef storeTree(inputTree,filename):\n    import pickle\n    fw = open(filename,&#039;w&#039;)\n    pickle.dump(inputTree,fw)\n    fw.close\ndef grabTree(filename):\n    import pickle\n    fr = open(filename)\n    return pickle.load(fr)\n<\/code><\/pre>\n<p>\u4e4b\u540e\u5728test.py\u4e2d\u8fd0\u884c\u4e00\u6b21Tree.storeTree()\uff0c\u5c31\u53ef\u4ee5\u4f7f\u7528Tree.grabTree()\u8bfb\u53d6\u51b3\u7b56\u6811\u4e86<br \/>\n<a href=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-5.png\"><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-5.png\" alt=\"\" width=\"1170\" height=\"699\" class=\"alignnone size-full wp-image-339\" srcset=\"http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-5.png 1170w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-5-300x179.png 300w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-5-768x459.png 768w, http:\/\/www.mrtblog.cn\/wp-content\/uploads\/2017\/02\/Decision-Tree-example-5-1024x612.png 1024w\" sizes=\"auto, (max-width: 1170px) 100vw, 1170px\" \/><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>1,374 Views\u53c2\u8003\u4e66\u7c4d\u300a\u673a\u5668\u5b66\u4e60\u5b9e\u6218\u300b \u51b3\u7b56\u6811\u7b97\u6cd5\u6982\u8ff0 K-\u8fd1\u90bb\u7b97\u6cd5\u53ef\u4ee5\u5b8c\u6210\u5f88\u591a\u5206\u7c7b\u4efb\u52a1\uff0c\u4f46\u662f\u5b83\u6700\u5927\u7684 [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[105],"tags":[43,47],"class_list":["post-285","post","type-post","status-publish","format-standard","hentry","category-machine-learning","tag-python","tag-47"],"_links":{"self":[{"href":"http:\/\/www.mrtblog.cn\/index.php?rest_route=\/wp\/v2\/posts\/285","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.mrtblog.cn\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.mrtblog.cn\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.mrtblog.cn\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.mrtblog.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=285"}],"version-history":[{"count":11,"href":"http:\/\/www.mrtblog.cn\/index.php?rest_route=\/wp\/v2\/posts\/285\/revisions"}],"predecessor-version":[{"id":1191,"href":"http:\/\/www.mrtblog.cn\/index.php?rest_route=\/wp\/v2\/posts\/285\/revisions\/1191"}],"wp:attachment":[{"href":"http:\/\/www.mrtblog.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=285"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.mrtblog.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=285"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.mrtblog.cn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=285"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}