def FromAltCrtTable(altCrtMat, crtName):
epsilon = 1e-307
crtNames = altCrtMat.CriterionNames()
altNames = altCrtMat.AlternativeNames()
crtIdx = crtNames.index(crtName)
if crtName not in crtNames:
return None
cmpTbl = AltCmpTable(crtName, altNames)
cmpMat = Matrix((len(altNames), len(altNames)))
valList = [
altCrtMat.AltCrtValue(altNames.index(altName), crtIdx)
for altName in altNames
]
for rowIdx in range(len(altNames)):
for colIdx in range(rowIdx, len(altNames)):
cmpMat[rowIdx][colIdx] = (valList[rowIdx] + epsilon) / (
valList[colIdx] + epsilon)
cmpMat[colIdx][rowIdx] = 1.0 / (cmpMat[rowIdx][colIdx] +
epsilon)
cmpTbl.SetComparisonMatrix(cmpMat.arr2d_)
cmpTbl.CalculateProperties()
return cmpTbl
def _LoadFromDomTree(self, domRoot): root = domRoot.getElementsByTagName(self.altcrtRootTag)[0] criteriaNamesNode = root.getElementsByTagName(self.criterionNamesTag)[0] #解析评价指标名称 self.crtNames_ = [self._GetText(criterionNameNode.childNodes) for criterionNameNode in criteriaNamesNode.getElementsByTagName(self.criterionNameTag)] evalCount = len(self.crtNames_) #解析评价目标-评价指标矩阵 altMatrixNode = root.getElementsByTagName(self.matrixTag)[0] altNodes = altMatrixNode.getElementsByTagName(self.altTag) altCount = len(altNodes) self.priorities_ = [0.0 for i in range(altCount)] self.mat_ = Matrix((altCount, evalCount)) self.altNames_ = ["" for i in range(altCount)] #读取矩阵 i_row = 0 for altNode in altNodes: self.altNames_[i_row] = altNode.getAttribute(self.altNameAttrTag) #读取alternative对应每个criteria的值 for altcrtFactorNode in altNode.getElementsByTagName(self.altcrtFactorTag): factor = float(self._GetText(altcrtFactorNode.childNodes)) altcrtFactorName = altcrtFactorNode.getAttribute(self.altcrtFactorNameAttrTag) i_col = self.crtNames_.index(altcrtFactorName) self.mat_[i_row][i_col] = factor i_row += 1
def __init__(self):
self.altNames_ = []
self.crtNames_ = []
self.mat_ = Matrix((0, 0))
self.priorities_ = []
self.criteria_ = []
self.rank2idxDict_ = {}
self.idx2rankDict_ = {}
return
def RemoveItem(self, itemName): try: idx = self.items_.index(itemName) except: return oldSize = len(self.items_) newSize = oldSize - 1 self.items_.remove(itemName) if newSize == 0: return oldSubMatLT = self.mat_.GetSub(0, 0, (idx, idx)) oldSubMatRT = self.mat_.GetSub(0, idx+1, (idx, oldSize-idx-1)) oldSubMatLB = self.mat_.GetSub(idx+1, 0, (oldSize-idx-1, idx)) oldSubMatRB = self.mat_.GetSub(idx+1, idx+1, (oldSize-idx-1, oldSize-idx-1)) self.mat_ = Matrix((newSize, newSize)) self.mat_.SetSub(0, 0, oldSubMatLT) self.mat_.SetSub(0, idx, oldSubMatRT) self.mat_.SetSub(idx, 0, oldSubMatLB) self.mat_.SetSub(idx, idx, oldSubMatRB)
def __init__(self, items, RIDict = None): """ 描述：初始化函数。 参数： items：AHP矩阵的要素名称列表 RIList: 随即一致性指标列表字典 """ self.items_ = items matSize = len(items) self.mat_ = Matrix((matSize, matSize)) self.propNames_ = ['MaxEigenvalue', 'CI', 'RI', 'CR', 'Consistancy'] self.RIDict_ = RIDict if self.RIDict_ == None: RIList = [ 0.00, 0.00, 0.58, 0.90, 1.12, #5个一组 1.24, 1.32, 1.41, 1.46, 1.49, 1.52, 1.54, 1.56, 1.58, 1.59 ] self.RIDict_ = dict([(i+1, RIList[i]) for i in range(len(RIList))]) self._InitializeProperties() return
class AHPComparisonTable: u""" 评价要素对评价目标的影响表。 """ def __init__(self, items, RIDict = None): """ 描述：初始化函数。 参数： items：AHP矩阵的要素名称列表 RIList: 随即一致性指标列表字典 """ self.items_ = items matSize = len(items) self.mat_ = Matrix((matSize, matSize)) self.propNames_ = ['MaxEigenvalue', 'CI', 'RI', 'CR', 'Consistancy'] self.RIDict_ = RIDict if self.RIDict_ == None: RIList = [ 0.00, 0.00, 0.58, 0.90, 1.12, #5个一组 1.24, 1.32, 1.41, 1.46, 1.49, 1.52, 1.54, 1.56, 1.58, 1.59 ] self.RIDict_ = dict([(i+1, RIList[i]) for i in range(len(RIList))]) self._InitializeProperties() return def _InitializeProperties(self): self.weights_ = [] self.ranks_ = [] self.maxLambda_ = 0.0 self.CI_ = 0.0 self.RI_ = 0.0 self.CR_ = 0.0 self.isConsistancy_ = False def _UpdateByLTri(self): size = self.mat_.Size() if size[0] == 0: return for iRow in range(1, size[0]): for iCol in range(iRow): if abs(self.mat_[iRow][iCol] < 1e-16): self.mat_[iCol][iRow] = 0.0 else: self.mat_[iCol][iRow] = 1.0 / self.mat_[iRow][iCol] for iElem in range(0, size[0]): self.mat_[iElem][iElem] = 1.0 def _UpdateByUTri(self): size = self.mat_.Size() if size[0] == 0: return for iCol in range(1, size[0]): for iRow in range(iCol): if abs(self.mat_[iRow][iCol] < 1e-16): self.mat_[iCol][iRow] = 0.0 else: self.mat_[iCol][iRow] = 1.0 / self.mat_[iRow][iCol] for iElem in range(0, size[0]): self.mat_[iElem][iElem] = 1.0 def _MaxLambda(self): self.maxLambda_, self.weights_ = MaxEigenvalue(self.mat_) def _ComputeConsistancyProperties(self): matSize = len(self.items_) self.CI_ = 0.0 self.RI_ = 0.0 self.CR_ = 0.0 if matSize > 2: self.CI_ = (self.maxLambda_ - matSize) / (matSize - 1) if matSize in self.RIDict_: self.RI_ = self.RIDict_[matSize] self.CR_ = self.CI_ / self.RI_ else: self.RI_ = 1.60 self.CR_ = self.CI_ / self.RI_ self.isConsistancy_ = (self.CR_ <= 0.1) def _RankItems(self): """ 根据权重对项目进行排名 """ if not (self.weights_ and len(self.weights_) > 0): return #将权重与标号对应，并按照权重进行排名。 #得到排名后的标号-权重列表 sortedWeightList = [(idx, self.weights_[idx]) for idx in range(len(self.weights_))] sortedWeightList.sort(cmp=lambda lhs, rhs: cmp(rhs[1], lhs[1])) #得到排序后的标号列表 idxSeq = [item[0] for item in sortedWeightList] #查找标号i在序列中的位置，便可得到该标号的rank self.ranks_ = [idxSeq.index(idx)+1 for idx in range(len(idxSeq))] def ComparisonMatrix(self): return self.mat_ def SetComparisonMatrix(self, arr2d): if Matrix.GetArraySize(arr2d) == self.mat_.Size(): self.mat_ = Matrix.FromArray2D(arr2d) def GetItemNames(self): return self.items_ def GetMatrixSize(self): return self.mat_.Size() def AddItem(self, itemName): self.items_.append(itemName) oldMat = self.mat_ self.mat_ = Matrix((len(self.items_), len(self.items_))) self.mat_.SetSub(0, 0, oldMat) return def RemoveItem(self, itemName): try: idx = self.items_.index(itemName) except: return oldSize = len(self.items_) newSize = oldSize - 1 self.items_.remove(itemName) if newSize == 0: return oldSubMatLT = self.mat_.GetSub(0, 0, (idx, idx)) oldSubMatRT = self.mat_.GetSub(0, idx+1, (idx, oldSize-idx-1)) oldSubMatLB = self.mat_.GetSub(idx+1, 0, (oldSize-idx-1, idx)) oldSubMatRB = self.mat_.GetSub(idx+1, idx+1, (oldSize-idx-1, oldSize-idx-1)) self.mat_ = Matrix((newSize, newSize)) self.mat_.SetSub(0, 0, oldSubMatLT) self.mat_.SetSub(0, idx, oldSubMatRT) self.mat_.SetSub(idx, 0, oldSubMatLB) self.mat_.SetSub(idx, idx, oldSubMatRB) def RenameItem(self, oldName, newName): oldIndex = self.items_.index(oldName) self.items_[oldIndex] = newName def SwapItem(self, lhsItemName, rhsItemName): try: lhsId = self.items_.index(lhsItemName) rhsId = self.items_.index(rhsItemName) if lhsId == rhsId: return False except: return False #Swap Matrix lhsRow = self.mat_[lhsId] lhsCol = self.mat_.GetCol(lhsId) rhsRow = self.mat_[rhsId] rhsCol = self.mat_.GetCol(rhsId) self.mat_[lhsId] = rhsRow self.mat_[rhsId] = lhsRow self.mat_.SetCol(lhsId, rhsCol) self.mat_.SetCol(rhsId, lhsCol) #Swap Items self.items_[rhsId] = lhsItemName self.items_[lhsId]= rhsItemName return True def CalculateProperties(self): try: self._MaxLambda() self._ComputeConsistancyProperties() self._RankItems() except: traceback.print_exc() self._InitializeProperties() def Weight(self, itemName): idx = self.items_.index(itemName) if idx >= len(self.weights_): self.weights_ = [0.0 for i in range(idx+1)] return self.weights_[idx] def Rank(self, itemName): idx = self.items_.index(itemName) if idx >= len(self.ranks_): self.ranks_ = [0 for i in range(idx+1)] return self.ranks_[idx] def MaxEigenvalue(self): return self.maxLambda_ def CI(self): return self.CI_ def RI(self): return self.RI_ def CR(self): return self.CR_ def Consistancy(self): return self.isConsistancy_ def GetPropertyNames(self): return self.propNames_ def GetProperty(self, propertyName): return getattr(self ,propertyName)()
def AddItem(self, itemName): self.items_.append(itemName) oldMat = self.mat_ self.mat_ = Matrix((len(self.items_), len(self.items_))) self.mat_.SetSub(0, 0, oldMat) return