diff --git a/src/python/09.RegTrees/regTrees.py b/src/python/09.RegTrees/regTrees.py index c6077e0f..b621c1cb 100644 --- a/src/python/09.RegTrees/regTrees.py +++ b/src/python/09.RegTrees/regTrees.py @@ -103,6 +103,7 @@ def chooseBestSplit(dataSet, leafType=regLeaf, errType=regErr, ops=(1, 4)): bestS, bestIndex, bestValue = inf, 0, 0 # 循环处理每一列对应的feature值 for featIndex in range(n-1): + # [0]表示这一列的[所有行]，不要[0]就是一个array[[所有行]] for splitVal in set(dataSet[:, featIndex].T.tolist()[0]): # 对该列进行分组，然后组内的成员的val值进行 二元切分 mat0, mat1 = binSplitDataSet(dataSet, featIndex, splitVal) @@ -236,7 +237,7 @@ def linearSolve(dataSet): # 如果矩阵的逆不存在，会造成程序异常 if linalg.det(xTx) == 0.0: raise NameError('This matrix is singular, cannot do inverse,\ntry increasing the second value of ops') - # 最小二乘法求最优解 + # 最小二乘法求最优解: w0*1+w1*x1=y ws = xTx.I * (X.T * Y) return ws, X, Y @@ -291,7 +292,9 @@ if __name__ == "__main__": # # 回归树 # myDat = loadDataSet('testData/RT_data1.txt') # # myDat = loadDataSet('testData/RT_data2.txt') + # # print 'myDat=', myDat # myMat = mat(myDat) + # # print 'myMat=', myMat # myTree = createTree(myMat) # print myTree @@ -301,7 +304,7 @@ if __name__ == "__main__": # myTree = createTree(myMat, ops=(0, 1)) # print myTree - # # 2.后剪枝就是：通过测试数据，对预测模型进行合并判断 + # # 2. 后剪枝就是：通过测试数据，对预测模型进行合并判断 # myDatTest = loadDataSet('testData/RT_data3test.txt') # myMat2Test = mat(myDatTest) # myFinalTree = prune(myTree, myMat2Test) @@ -330,11 +333,11 @@ if __name__ == "__main__": print myTree2 print "模型树:", corrcoef(yHat2, testMat[:, 1],rowvar=0)[0, 1] - # # 线性回归 - # ws, X, Y = linearSolve(trainMat) - # print ws - # m = len(testMat[:, 0]) - # yHat3 = mat(zeros((m, 1))) - # for i in range(shape(testMat)[0]): - # yHat3[i] = testMat[i, 0]*ws[1, 0] + ws[0, 0] - # print "线性回归:", corrcoef(yHat3, testMat[:, 1],rowvar=0)[0, 1] + # 线性回归 + ws, X, Y = linearSolve(trainMat) + print ws + m = len(testMat[:, 0]) + yHat3 = mat(zeros((m, 1))) + for i in range(shape(testMat)[0]): + yHat3[i] = testMat[i, 0]*ws[1, 0] + ws[0, 0] + print "线性回归:", corrcoef(yHat3, testMat[:, 1],rowvar=0)[0, 1] diff --git a/src/python/09.RegTrees/treeExplore.py b/src/python/09.RegTrees/treeExplore.py index a426a342..33665a6a 100644 --- a/src/python/09.RegTrees/treeExplore.py +++ b/src/python/09.RegTrees/treeExplore.py @@ -100,7 +100,6 @@ def main(root): # 退出按钮 Button(root, text="退出", fg="black", command=quit).grid(row=1, column=2) - # 创建一个画板 canvas reDraw.f = Figure(figsize=(5, 4), dpi=100) reDraw.canvas = FigureCanvasTkAgg(reDraw.f, master=root) diff --git a/src/python/11.Apriori/apriori.py b/src/python/11.Apriori/apriori.py index 3ba163d0..b4daef2b 100644 --- a/src/python/11.Apriori/apriori.py +++ b/src/python/11.Apriori/apriori.py @@ -43,7 +43,7 @@ def scanD(D, Ck, minSupport): Args: D 原始数据集, D用来判断，CK中的元素，是否存在于原数据D中 - Ck 合并后的数据集 + Ck 所有key的元素集合 Returns: retList 支持度大于阈值的集合 supportData 全量key的字典集合 @@ -141,6 +141,8 @@ def apriori(dataSet, minSupport=0.5): if len(Lk) == 0: break # Lk表示满足频繁子项的集合，L元素在增加 + # l=[[set(1), set(2), set(3)]] + # l=[[set(1), set(2), set(3)] [set(1, 2), set(2, 3)]] L.append(Lk) k += 1 # print 'k=', k, len(L[k-2]) @@ -157,7 +159,7 @@ def calcConf(freqSet, H, supportData, brl, minConf=0.7): brl bigRuleList的空数组 minConf 置信度的阈值 Returns: - prunedH 记录 可信度大于阈值的集合 + prunedH 记录 置信度大于阈值的集合 """ # 记录 可信度大于阈值的集合 prunedH = [] @@ -188,7 +190,7 @@ def rulesFromConseq(freqSet, H, supportData, brl, minConf=0.7): """ # H[0]是freqSet的元素组合的第一个元素 m = len(H[0]) - # 判断，freqSet的长度是否>组合的长度+1, 避免过度匹配 例如：计算过一边{1,2,3} 和 {1, 2} {1, 3}，就没必要再计算了 {1,2,3}和{1,2,3}的组合关系 + # 判断，freqSet的长度是否>组合的长度+1, 避免过度匹配 例如：计算过一边{1,2,3} 和 {1, 2} {1, 3}，就没必要再计算了进一步合并来计算 {1,2,3}和{1,2,3}的组合关系 if (len(freqSet) > (m + 1)): print 'freqSet******************', len(freqSet), m + 1, freqSet, H, H[0] # 合并数据集集合，组合为2/3/..n的集合 @@ -209,7 +211,7 @@ def generateRules(L, supportData, minConf=0.7): Args: L 频繁项集的全集 supportData 所有元素和支持度的全集 - minConf 可信度的阈值 + minConf 置信度的阈值 Returns: bigRuleList 关于 (A->B+置信度) 3个字段的组合 """ @@ -217,7 +219,9 @@ def generateRules(L, supportData, minConf=0.7): # 循环L频繁项集，所有的统一大小组合（2/../n个的组合，从第2组开始） for i in range(1, len(L)): # 获取频繁项集中每个组合的所有元素 + # [[frozenset([1]), frozenset([3]), frozenset([2]), frozenset([5])], [frozenset([1, 3]), frozenset([2, 5]), frozenset([2, 3]), frozenset([3, 5])], [frozenset([2, 3, 5])]] for freqSet in L[i]: + # 假设：freqSet=frozenset([1, 3]) H1=[1, 3] # 组合总的元素并遍历子元素，并转化为冻结的set集合，再存放到list列表中 H1 = [frozenset([item]) for item in freqSet] # 2个的组合，走else, 2个以上的组合，走if diff --git a/src/python/12.FrequentPattemTree/fpGrowth.py b/src/python/12.FrequentPattemTree/fpGrowth.py index 46639ca0..42a9f3f7 100644 --- a/src/python/12.FrequentPattemTree/fpGrowth.py +++ b/src/python/12.FrequentPattemTree/fpGrowth.py @@ -41,6 +41,7 @@ def loadSimpDat(): ['z', 'y', 'x', 'w', 'v', 'u', 't', 's'], ['z'], ['r', 'x', 'n', 'o', 's'], + # ['r', 'x', 'n', 'o', 's'], ['y', 'r', 'x', 'z', 'q', 't', 'p'], ['y', 'z', 'x', 'e', 'q', 's', 't', 'm']] return simpDat @@ -49,7 +50,10 @@ def loadSimpDat(): def createInitSet(dataSet): retDict = {} for trans in dataSet: - retDict[frozenset(trans)] = 1 + if not retDict.has_key(frozenset(trans)): + retDict[frozenset(trans)] = 1 + else: + retDict[frozenset(trans)] += 1 return retDict @@ -193,7 +197,7 @@ def findPrefixPath(basePat, treeNode): # prefixPath[1:] 变frozenset后，字母就变无序了 # condPats[frozenset(prefixPath)] = treeNode.count condPats[frozenset(prefixPath[1:])] = treeNode.count - # 递归，寻找改节点的上一个 相同值的链接节点 + # 递归，寻找改节点的下一个 相同值的链接节点 treeNode = treeNode.nodeLink # print treeNode return condPats