灰色关联分析计算的实例演示1.根据评价目的确定评价指标体系,收集评价数据。1.灰色关联分析计算步骤设n个数据序列形成如下矩阵:mxmxmxxxxxxxXXXnnnn21212121222111,,其中为指标的个数,.mnimxxxXTiiii,,2,1,,,2,12.确定参考数据列参考数据列应该是一个理想的比较标准,可以以各指标的最优值(或最劣值)构成参考数据列,也可根据评价目的选择其它参照值.记作mxxxX0000,,2,)1(3.对指标数据进行无量纲化无量纲化后的数据序列形成如下矩阵:mxmxmxxxxxxxXXXnnnn10101010222111,,,常用的无量纲化方法有均值化法(见(12-3)式)、初值化法(见(12-4)式)和变换等.sxx.,,2,1,,1,0)412(1)312(11mknixkxkxkxmkxkxiiimkiii;4.逐个计算每个被评价对象指标序列(比较序列)与参考序列对应元素的绝对差值即(,为被评价对象的个数).5.确定与)()(0kxkximk,,1ni,,1n)()(minmin011kxkximkni)()(maxmax011kxkximkni6.计算关联系数由(12-5)式,分别计算每个比较序列与参考序列对应元素的关联系数.0000minmin()()maxmax()()()125)()()maxmax()()iiikikiiiikxkxkxkxkkxkxkxkxk(mk,,10.5式中为分辨系数,在(0,1)内取值,若越小,关联系数间差异越大,区分能力越强。通常取7.计算关联度对各评价对象(比较序列)分别计算其个指标与参考序列对应元素的关联系数的均值,以反映各评价对象与参考序列的关联关系,并称其为关联度,记为:2.自然灾害经济损失及相关因素灰色关联分析根据灰色关联分析中关联系数和关联度的计算公式,利用自然灾害经济损失的2000-2004年有关原始数据(见表1),计算了灾害直接经济损失(参考序列)同形成灾害经济损失的各因素(比较序列)之间的关联度(见表2)。1.建立原始数据矩阵:2045.31942.21637.21884.21602.33437431793273193251616297(X´)=14.679214.84491.477446.6049.4959120.9100.165.980.5254.220.30690.74090.3613.72.021349.420134.869950.97450.432540.8828年份20002001200220032004灾害直接经济损失(亿元)2045.31942.21637.21884.21602.3农作物成灾面积(千公顷)3437431793273193251616297地震灾害损失(亿元)14.679214.84491.477446.6049.4959海洋灾害损失(亿元)120.9100.165.980.5254.22森林火灾损失(亿元)0.30690.74090.3613.72.0213地质灾害损失(亿元)49.420134.869950.97450.432540.88282.矩阵无量纲化(初值化):X=Xij´/Xi1´(i=1,2,3,4,5,6;j=2,3,4,5)10.94960.80050.92120.783410.92490.79480.94590.4741(X)=11.01130.10063.17480.646910.82800.54510.66600.448512.41411.176312.05606.586210.70561.03141.02050.82733.确定参考数据列:X0=1,0.9496,0.8005,0.9212,0.78344.计算|X0-Xi|:1=(0,0.0247,0.0057,0.0247,0.3093)2=(0,0.0617,0.6998,2.2536,0.1365)3=(0,0.1216,0.2554,0.2552,0.3349)4=(0,1.4645,0.3758,11.1348,5.8028)5=(0,0.2440,0.2310,0.0993,0.0438)00.02470.00570.02470.309300.06170.69982.25360.1365()=00.12160.25540.25520.334901.46450.375811.13485.802800.24400.23100.09930.04385.求最值:=0=11.13486.计算关联系数(以矩阵第一行为例):ξ(0)=(0+11.1348*0.5)/(0+11.1348*0.5)=1ξ(0.0247)=(0+11.1348*0.5)/(0.0247+11.1348*0.5)=0.9956ξ(0.0057)=(0+11.1348*0.5)/(0.0057+11.1348*0.5)=0.9990ξ(0.0247)=(0+11.1348*0.5)/(0.0247+11.1348*0.5)=0.9956ξ(0.3093)=(0+11.1348*0.5)/(0.3093+11.1348*0.5)=0.9474网赚)()(minmin011kxkximkni)()(maxmax011kxkximkni编号ξξξξξ110.99560.99900.99560.9474210.98900.88830.71190.9761310.97860.95610.95620.9433410.79170.93680.33330.4896510.95800.96020.98250.9922同理可得其它值,见下表:7.分别计算每个指标的关联度:r1=(1+0.9956+0.9990+0.9956+0.9474)/5=0.9875r2=(1+0.9890+0.8883+0.7119+0.9761)/5=0.9131r3=(1+0.9786+0.9561+0.9562+0.9433)/5=0.9668r4=(1+0.9580+0.9602+0.9825+0.9922)/5=0.7103r5=(1+0.9580+0.9602+0.9825+0.9922)/5=0.9786由表2的结果可以看出,灾害经济损失的各相关影响因素对灾害直接经济损失影响的关联度大小的顺序为:农作物成灾面积地质灾害损失海洋灾害损失地震灾害损失森林火灾损失可以说明对灾害直接经济损失影响最大的是农作物成灾面积、地质灾害损失和海洋灾害损失,其次为地震灾害损失,森林火灾损失对灾害直接经济损失影响程度较小。