{"id":128,"date":"2018-06-08T21:26:08","date_gmt":"2018-06-08T13:26:08","guid":{"rendered":"http:\/\/www.sniper97.cn\/?p=128"},"modified":"2018-06-08T21:26:08","modified_gmt":"2018-06-08T13:26:08","slug":"knn%e7%ae%97%e6%b3%95","status":"publish","type":"post","link":"http:\/\/www.sniper97.cn\/index.php\/note\/machine-learning-in-action\/128\/","title":{"rendered":"KNN\u7b97\u6cd5"},"content":{"rendered":"<h1><strong>K-\u8fd1\u90bb\u7b97\u6cd5\uff08<em>k<\/em>-Nearest Neighbor\uff09<\/strong><\/h1>\n<p>\u7b80\u5355\u7684\u8bf4\uff0cK-\u8fd1\u90bb\u7b97\u6cd5\u91c7\u7528\u6d4b\u91cf\u4e0d\u540c\u7279\u5f81\u503c\u4e4b\u95f4\u7684\u8ddd\u79bb\u65b9\u6cd5\u8fdb\u884c\u5206\u7c7b\u3002\uff5b\u4f7f\u7528\u6b27\u51e0\u91cc\u5f97\u5ea6\u91cf\uff08euclidean metric\uff09\uff5d<br \/>\n\u4f18\u70b9\uff1a\u7cbe\u5ea6\u9ad8\u3001\u5bf9\u5f02\u5e38\u503c\u4e0d\u654f\u611f\u3001\u65e0\u6570\u636e\u8f93\u5165\u5047\u5b9a\u3002<br \/>\n\u7f3a\u70b9\uff1a\u8ba1\u7b97\u590d\u6742\u5ea6\u9ad8\u3001\u7a7a\u95f4\u590d\u6742\u5ea6\u9ad8\u3002<br \/>\n\u9002\u7528\u6570\u636e\u8303\u56f4\uff1a\u6570\u503c\u578b\u548c\u6807\u7a0b\u79f0\u578b\u3002<br \/>\n&nbsp;<br \/>\nK-\u8fd1\u90bb\u7b97\u6cd5\u7684\u4e00\u822c\u6d41\u7a0b<br \/>\n\uff081\uff09\u6536\u96c6\u6570\u636e\uff1a\u53ef\u4ee5\u4f7f\u7528\u4efb\u4f55\u65b9\u6cd5\u3002<br \/>\n\uff082\uff09\u51c6\u5907\u6570\u636e\uff1a\u8ddd\u79bb\u8ba1\u7b97\u6240\u9700\u8981\u7684\u6570\u503c\uff0c\u6700\u597d\u662f\u7ed3\u6784\u5316\u7684\u6570\u636e\u683c\u5f0f\u3002<br \/>\n\uff083\uff09\u5206\u6790\u6570\u636e\uff1a\u53ef\u4ee5\u4f7f\u7528\u4efb\u4f55\u65b9\u6cd5\u3002<br \/>\n\uff084\uff09\u8bad\u7ec3\u7b97\u6cd5\uff1a\u6b64\u6b65\u9aa4\u4e0d\u9002\u7528\u4e8eK-\u8fd1\u90bb\u7b97\u6cd5\u3002<br \/>\n\uff085\uff09\u6d4b\u8bd5\u7b97\u6cd5\uff1a\u8ba1\u7b97\u9519\u8bef\u7387\u3002<br \/>\n\uff086\uff09\u4f7f\u7528\u7b97\u6cd5\uff1a\u9996\u5148\u9700\u8981\u8f93\u5165\u6837\u672c\u6570\u636e\u548c\u7ed3\u6784\u5316\u7684\u8f93\u51fa\u7ed3\u679c\uff0c\u7136\u540e\u8fd0\u884cK-\u8fd1\u90bb\u7b97\u6cd5\u5224\u65ad\u8f93\u5165\u6570\u636e\u5206\u522b\u5c5e\u4e8e\u54ea\u4e2a\u5206\u7c7b\uff0c\u6700\u540e\u5e94\u7528\u5bf9\u8ba1\u7b97\u51fa\u7684\u5206\u7c7b\u6267\u884c\u540e\u7eed\u7684\u5904\u7406\u3002<br \/>\n&nbsp;<br \/>\nK-\u8fd1\u90bb\u7b97\u6cd5\u662f\u5206\u7c7b\u6570\u636e\u6700\u7b80\u5355\u6700\u6709\u6548\u7684\u7b97\u6cd5\u3002K-\u8fd1\u90bb\u7b97\u6cd5\u662f\u57fa\u4e8e\u5b9e\u4f8b\u7684\u5b66\u4e60\uff0c\u4f7f\u7528\u7b97\u6cd5\u65f6\u6211\u4eec\u5fc5\u987b\u6709\u63a5\u8fd1\u5b9e\u9645\u6570\u636e\u7684\u8bad\u7ec3\u6837\u672c\u6570\u636e\u3002K-\u8fd1\u90bb\u7b97\u6cd5\u5fc5\u987b\u4fdd\u5b58\u5168\u90e8\u6570\u636e\u96c6\uff0c\u5982\u679c\u8bad\u7ec3\u6570\u636e\u96c6\u7684\u5f88\u5927\uff0c\u5fc5\u987b\u4f7f\u7528\u5927\u91cf\u7684\u5b58\u50a8\u7a7a\u95f4\u3002\u6b64\u5916\uff0c\u7531\u4e8e\u5fc5\u987b\u5bf9\u6570\u636e\u96c6\u4e2d\u7684\u6ca1\u4e2a\u6570\u636e\u8ba1\u7b97\u8ddd\u79bb\u503c\u3002\u5b9e\u9645\u4f7f\u7528\u65f6\u53ef\u80fd\u975e\u5e38\u8017\u65f6\u3002<br \/>\nK-\u8fd1\u90bb\u7b97\u6cd5\u7684\u53e6\u4e00\u4e2a\u7f3a\u9677\u662f\u65e0\u6cd5\u7ed9\u51fa\u4efb\u4f55\u6570\u636e\u7684\u57fa\u7840\u7ed3\u6784\u4fe1\u606f\uff0c\u56e0\u6b64\u6211\u4eec\u4e5f\u65e0\u6cd5\u77e5\u6653\u5e73\u5747\u5b9e\u4f8b\u6837\u672c\u548c\u5178\u578b\u5b9e\u4f8b\u6837\u672c\u5177\u6709\u4ec0\u4e48\u7279\u5f81\u3002<br \/>\n&nbsp;<br \/>\n\u5b98\u65b9\u7ed9\u7684\u662fpython 2\u7684\u4ee3\u7801\uff0c\u5f88\u591a\u4e1c\u897f\u5728python 3\u4e2d\u5df2\u7ecf\u65e0\u6cd5\u4f7f\u7528\uff0c\u5e76\u4e14python 2 \u4e5f\u5feb\u505c\u6b62\u7ef4\u62a4\u4e86\u3002<br \/>\n\u589e\u52a0\u4e86\u5f88\u591a\u6ce8\u91ca\uff0c\u80fd\u591f\u66f4\u65b9\u4fbf\u7684\u7406\u89e3 \u3002<br \/>\n&nbsp;<\/p>\n<pre class=\"EnlighterJSRAW\" data-enlighter-language=\"python\"># -*- coding: utf-8 -*-\n# K\u9636\u90bb\u7b97\u6cd5\nfrom numpy import *\nimport operator\nimport matplotlib\nimport matplotlib.pyplot as plt\nfrom os import listdir\ndef createDataSet():\n    group = array([[1.0, 1.1], [1.0, 1.0], [0, 0], [0, 0.1]])\n    labels = ['A', 'A', 'B', 'B']\n    return group, labels\n# \u7528kNN\u8ba1\u7b97\u7535\u5f71\u7ea7\u522b\ndef classify0(inX, dataSet, labels, k):  # \u6d4b\u8bd5  \u8bad\u7ec3  \u7ed3\u679c  k\u9636\n    # shape \u7684\u529f\u80fd\u662f\u67e5\u770b\u77e9\u9635\u6216\u8005\u6570\u7ec4\u7684\u7ef4\u6570\u3002\n    # shape()\u6216\u8005a.shape\u8fd4\u56de\u7684\u662f\u77e9\u9635\u7684\u884c\u5217\u6570\uff08[a,b]\uff0c\u5176\u4e2da\u662f\u884c\u6570\uff0cb\u662f\u5217\u6570\uff09\n    # shape[0] \u53ea\u8fd4\u56de\u884c\u6570  shape[1] \u53ea\u8fd4\u56de\u5217\u6570\n    dataSetSize = dataSet.shape[0]\n    # tile \u91cd\u590dA\u7684\u5404\u4e2a\u7ef4\u5ea6\n    # tile\uff08A,B\uff09 \u5728\u884c\u65b9\u5411\u4e0a\u91cd\u590dA\u6b21\uff0c\u5728\u5217\u65b9\u5411\u4e0a\u91cd\u590dB\u6b21\n    diffMat = tile(inX, (dataSetSize, 1)) - dataSet\n    # diffMat\u7684\u4e8c\u6b21\u65b9  \uff08a**b == a\u7684b\u6b21\u65b9\uff09\n    sqDiffMat = diffMat ** 2\n    # sum  numpy \u4e0b\u7684\u65b9\u6cd5 \u9ed8\u8ba4axis\u4e3aNone\uff0c\u8868\u793a\u5c06\u6240\u6709\u5143\u7d20\u7684\u503c\u76f8\u52a0 axis=1\u8868\u793a\u6309\u884c\u76f8\u52a0 , axis=0\u8868\u793a\u6309\u5217\u76f8\u52a0\n    sqDistances = sqDiffMat.sum(axis=1)\n    # \u5f00\u6839\u53f7\n    distances = sqDistances ** 0.5\n    # argsort\u8fd4\u56de\u6570\u503c\u4ece\u5c0f\u5230\u5927\u7684\u7d22\u5f15\uff08\u503c\u6570\u7ec4\u7d22\u5f150,1,2,3\uff09\n    sortedDistIndecies = distances.argsort()\n    # \u65b0\u5efa\u5b57\u5178\n    classCount = {}\n    for i in range(k):\n        voteIlabel = labels[sortedDistIndecies[i]]\n        # \u5728\u5b57\u5178\u4e2d\u67e5\u627evoteIlabel\uff0c\u5982\u679c\u627e\u4e0d\u5230\u8fd4\u56de0\uff0c\u9ed8\u8ba4\u4e3a None\n        # \u4e0d\u65ad\u7d2f\u52a0\u8ba1\u6570\u7684\u8fc7\u7a0b\n        classCount[voteIlabel] = classCount.get(voteIlabel, 0) + 1\n    # python3\u4e2d\uff1aclassCount.iteritems()\u4fee\u6539\u4e3aclassCount.items()\n    # sorted(iterable, cmp=None, key=None, reverse=False) --&gt; new sorted list\u3002\n    # reverse\u9ed8\u8ba4\u5347\u5e8f\n    # key\u5173\u952e\u5b57\u6392\u5e8fitemgetter\uff081\uff09\u6309\u7167\u7b2c1\u4f4d\u5143\u7d20\u7684\u5927\u5c0f\uff08\u5b57\u5178\u5e8f\uff09\u6392\u5e8f\n    sortedClassCount = sorted(classCount.items(), key=operator.itemgetter(1), reverse=True)\n    # \u8fd4\u56de\u7684sortedClassCount \u662f\u5bf9\u51fa\u73b0\u6b21\u6570\u7684\u6392\u5e8f \u8fd4\u56de\u6392\u5e8f\u7684\u7b2c\u4e00\u4e2a\u5143\u7d20\u7684\u7b2c\u4e00\u4e2a\u5b50\u5143\u7d20\n    # eg:[('A', 2), ('B', 1)]   \u8fd4\u56de  ('A', 2)  \u4e2d\u7684  \u7b2c0\u4e2a  'A'\n    return sortedClassCount[0][0]\n# \u4ece\u6587\u4ef6\u91cc\u8bfb\u53d6\u6570\u636e\ndef file2Matrix(filename):\n    fr = open(filename)\n    # readlines() \u65b9\u6cd5\u7528\u4e8e\u8bfb\u53d6\u6240\u6709\u884c(\u76f4\u5230\u7ed3\u675f\u7b26 EOF)\u5e76\u8fd4\u56de\u5217\u8868\n    arrryOlines = fr.readlines()\n    # \u83b7\u53d6\u6570\u636e\u884c\u6570\n    numberOfLines = len(arrryOlines)\n    # zeros \u521b\u5efa0\u77e9\u9635\n    returnMat = zeros((numberOfLines, 3))\n    classLabelVector = []\n    index = 0\n    for line in arrryOlines:\n        line = line.strip()\n        listFromLine = line.split('\\t')\n        # t[index, :]   ==   t[index: , index: ]  \u5bf9\u4e8e\u4e8c\u7ef4\u6570\u7ec4\u7684\u904d\u5386\n        # t[: ,index]   ==   t[: index, : index]  \u5bf9\u4e8e\u4e8c\u7ef4\u6570\u7ec4\u7684\u904d\u5386\n        returnMat[index, :] = listFromLine[0:3]\n        # -1\u7d22\u5f15\u8868\u793a\u6700\u540e\u4e00\u5217\u5143\u7d20,\u4f4dlabel\u4fe1\u606f\u5b58\u50a8\u5728classLabelVector\n        classLabelVector.append(int(listFromLine[-1]))\n        index += 1\n    return returnMat, classLabelVector\n# \u753b\u7279\u5f81\u56fe\ndef paintFigure(datingDataMat, datingLabels):\n    fig = plt.figure()\n    ax = fig.add_subplot(111)\n    # \u6ca1\u6837\u672c\u5206\u7c7b\u7684\u7279\u5f81\u56fe\n    # ax.scatter(datingDataMat[:, 1], datingDataMat[:, 2])\n    # \u6709\u6837\u672c\u5206\u7c7b\u7684\u7279\u5f81\u56fe\n    ax.scatter(datingDataMat[:, 1], datingDataMat[:, 2], 15.0 * array(datingLabels), 15.0 * array(datingLabels))\n    plt.show()\n# \u5f52\u4e00\u5316\u6570\u503c\u65b9\u6cd5\ndef autoNorm(dataSet):\n    # min(0)\u4e2d\u7684\u53c2\u65700 \u4f7f\u5f97\u51fd\u6570\u53ef\u4ee5\u4ece\u4ece\u5217\u4e2d\u9009\u53d6\u6700\u5c0f\u503c\uff0c\u800c\u4e0d\u662f\u9009\u53d6\u5f53\u524d\u884c\u7684\u6700\u5c0f\u503c\n    minVals = dataSet.min(0)\n    # max(0)\u4e2d\u7684\u53c2\u65700 \u4f7f\u5f97\u51fd\u6570\u53ef\u4ee5\u4ece\u4ece\u5217\u4e2d\u9009\u53d6\u6700\u5927\u503c\uff0c\u800c\u4e0d\u662f\u9009\u53d6\u5f53\u524d\u884c\u7684\u6700\u5927\u503c\n    maxVals = dataSet.max(0)\n    ranges = maxVals - minVals\n    # \u521b\u5efa\u6570\u636e\u957f\u5ea6\u5927\u5c0f\u7684\u6570\u7ec4\uff08shape \u8fd4\u56de\u7ef4\u6570[a,b]\uff09\n    normDataSet = zeros(shape(dataSet))\n    # shape[0] \u53ea\u8fd4\u56de\u884c\u6570  shape[1] \u53ea\u8fd4\u56de\u5217\u6570\n    m = dataSet.shape[0]\n    # \u8fd9\u91cc\u662f\u5bf9\u77e9\u9635\u8fdb\u884c\u8ba1\u7b97\uff0c\u800c\u4e0d\u662f\u5bf9\u5355\u4e00\u6570\u636e\u8fdb\u884c\u8ba1\u7b97\uff0c\u7528tile()\u5bf9minVals\u5c55\u5f00\u5230m\uff0c\uff08m,1\uff09 \u6cbf\u7740\u5404\u4e2a\u7ef4\u5ea6\u5ef6\u4f38\u7684\u6b21\u6570\n    normDataSet = dataSet - tile(minVals, (m, 1))\n    normDataSet = normDataSet \/ tile(ranges, (m, 1))\n    return normDataSet, ranges, minVals\n# \u5bf9\u6587\u4ef6\u6570\u636e\u8fdb\u884ck\u9636\u90bb\u7b97\u6cd5\ndef datingClassTest():\n    # \u6d4b\u8bd5\u6570\u636e\u767e\u5206\u6bd4\n    hoRatio = 0.10\n    datingDataMat, datingLabels = file2Matrix(\n        'G:\\MLiA_SourceCode\\machinelearninginaction\\Ch02\\datingTestSet2.txt')\n    normMat, ranges, minVals = autoNorm(datingDataMat)\n    m = normMat.shape[0]\n    # \u7528\u4e8e\u6d4b\u8bd5\u7684\u6570\u636e\u6570\u91cf\n    numTestVecs = int(m * hoRatio)\n    # \u9519\u8bef\u6570\u636e\u6570\n    errorcount = 0.0\n    for i in range(numTestVecs):\n        classifierResult = classify0(normMat[i, :], normMat[numTestVecs:m, :], datingLabels[numTestVecs:m], 3)\n        print(\"the classifer came back with:%d\uff0cthe real answer is:%d\" % (classifierResult, datingLabels[i]))\n        if (classifierResult != datingLabels[i]):\n            errorcount += 1\n    print(\"the total errror rate is:%f\" % (errorcount \/ float(numTestVecs)))\ndef classifyPerson():\n    resultList = ['not at all', 'in small doses', 'in large doses']\n    percentTats = float(input(\"percentage of time spent playing miles earned per year?\"))\n    ffMiles = float(input(\"frequent flier miles earned per year?\"))\n    iceCream = float(input(\"liters of ice cream consumed per year?\"))\n    datingDataMat, datingLabels = file2Matrix(\n        \"G:\\MLiA_SourceCode\\machinelearninginaction\\Ch02\\datingTestSet2.txt\")\n    normMat, ranges, minVals = autoNorm(datingDataMat)\n    inArr = array([ffMiles, percentTats, iceCream])\n    classifierResult = classify0((inArr - minVals) \/ ranges, normMat, datingLabels, 3)\n    print(\"You will probably like this person:\", resultList[classifierResult - 1])\n# \u5c06\u4e8c\u8fdb\u5236\u56fe\u7247\u5b58\u5230\u6570\u7ec4\u4e2d\ndef img2vector(filename):\n    returnVect = zeros((1, 1024))\n    fr = open(filename)\n    for i in range(32):\n        lineStr = fr.readline()\n        for j in range(32):\n            returnVect[0, 32 * i + j] = int(lineStr[j])\n        return returnVect\n# \u624b\u5199\u6570\u5b57\u7cfb\u7edf\u7684\u6d4b\u8bd5\u4ee3\u7801\ndef handwritingClassTest():\n    hwLabels = []\n    #\n    trainingFileList = listdir('G:\/MLiA_SourceCode\/machinelearninginaction\/Ch02\/digits\/trainingDigits')\n    m = len(trainingFileList)\n    trainingMat = zeros((m, 1024))\n    for i in range(m):\n        fileNameStr = trainingFileList[i]\n        fileStr = fileNameStr.split('.')[0]\n        classNumStr = int(fileStr.split('_')[0])\n        hwLabels.append(classNumStr)\n        trainingMat[i, :] = img2vector(\n            'G:\/MLiA_SourceCode\/machinelearninginaction\/Ch02\/digits\/trainingDigits\/%s' % fileNameStr)\n    testFileList = listdir('G:\/MLiA_SourceCode\/machinelearninginaction\/Ch02\/digits\/testDigits')\n    errorCount = 0.0\n    mTest = len(testFileList)\n    for i in range(mTest):\n        fileNameStr = testFileList[i]\n        fileStr = fileNameStr.split('.')[0]\n        classNumStr = int(fileStr.split('_')[0])\n        vectorUnderTest = img2vector(\n            'G:\/MLiA_SourceCode\/machinelearninginaction\/Ch02\/digits\/testDigits\/%s' % fileNameStr)\n        classifierResult = classify0(vectorUnderTest, trainingMat, hwLabels, 3)\n        print(\"the classifier came back with: %d,the real answer is: %d\" % (classifierResult, classNumStr))\n        if (classifierResult != classNumStr):\n            errorCount += 1.0\n    print(\"\\nthe total number of errors is: %d\" % errorCount)\n    print(\"\\nthe total error rate is: %f\" % (errorCount \/ float(mTest)))\nif __name__ == '__main__':\n    # \u6d4b\u8bd5\u7535\u5f71\u5206\u7ea7\u65b9\u6cd5\n    # groups, labels = createDataSet()\n    # print(classify0([0.5, 1.5], groups, labels, 3))\n    # \u6d4b\u8bd5\u753b\u7279\u5f81\u56fe\u65b9\u6cd5\n    # datingDataMat, datingLabels = file2Matrix(\n    #     'G:\\MLiA_SourceCode\\machinelearninginaction\\Ch02\\datingTestSet2.txt')\n    # paintFigure(datingDataMat, datingLabels)\n    # \u6d4b\u8bd5autoNorm \u5f52\u4e00\u5316\u6570\u503c\u65b9\u6cd5\n    # print(datingDataMat[0:5])\n    # print(\"\\n\\n\")\n    # datingDataMat, ranges, minVals = autoNorm(datingDataMat)\n    # print(ranges)\n    # print(\"\\n\\n\")\n    # print(minVals)\n    # print(\"\\n\\n\")\n    # print(datingDataMat[0:5])\n    # \u6d4b\u8bd5\u7ea6\u4f1a\u7f51\u7ad9\u8bc4\u7ea7\n    # datingClassTest()\n    # \u8f93\u5165\u6570\u636e\u6d4b\u8bd5\u7ea6\u4f1a\u7f51\u7ad9\n    # classifyPerson()\n    # \u6d4b\u8bd5\u624b\u5199\u7cfb\u7edf\n    handwritingClassTest()\n    # \u77e9\u9635\u8f93\u51fa\u6d4b\u8bd5\n    # a = arange(9)\n    # b = a.reshape(3, 3)\n    # print(b)\n    # print(\"\\n\\n\")\n    # print(b[0:1, 0:2])\n<\/pre>\n<p>&nbsp;<\/p>\n","protected":false},"excerpt":{"rendered":"<p>K-\u8fd1\u90bb\u7b97\u6cd5\uff08k-Nearest Neighbor\uff09 \u7b80\u5355\u7684\u8bf4\uff0cK-\u8fd1\u90bb\u7b97\u6cd5\u91c7\u7528\u6d4b\u91cf\u4e0d\u540c\u7279\u5f81\u503c\u4e4b\u95f4\u7684 [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"om_disable_all_campaigns":false,"_mi_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"footnotes":""},"categories":[6,10],"tags":[],"views":3106,"_links":{"self":[{"href":"http:\/\/www.sniper97.cn\/index.php\/wp-json\/wp\/v2\/posts\/128"}],"collection":[{"href":"http:\/\/www.sniper97.cn\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.sniper97.cn\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.sniper97.cn\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.sniper97.cn\/index.php\/wp-json\/wp\/v2\/comments?post=128"}],"version-history":[{"count":0,"href":"http:\/\/www.sniper97.cn\/index.php\/wp-json\/wp\/v2\/posts\/128\/revisions"}],"wp:attachment":[{"href":"http:\/\/www.sniper97.cn\/index.php\/wp-json\/wp\/v2\/media?parent=128"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.sniper97.cn\/index.php\/wp-json\/wp\/v2\/categories?post=128"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.sniper97.cn\/index.php\/wp-json\/wp\/v2\/tags?post=128"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}