1、 经济性指标经济性指标 社会经济系统常用的评价指标社会经济系统常用的评价指标 社会性指标社会性指标 技术性指标技术性指标 资源性指标资源性指标 政策性指标政策性指标 基础设施指标基础设施指标 其他指标其他指标产值、收入、成本、产值、收入、成本、税金、投资额、投税金、投资额、投资回收期、固定资资回收期、固定资产等等产等等人员素质、社会福人员素质、社会福利、生态环境、就利、生态环境、就业机会等业机会等产品性能、可靠性、产品性能、可靠性、工艺水平、人员素工艺水平、人员素质等质等矿产资源、水源、矿产资源、水源、土地、人力等土地、人力等国家和地方的政国家和地方的政策、法令、计划策、法令、计划等等交通、供
2、水、供交通、供水、供电等电等特定决策系统的特定决策系统的特有指标,如净特有指标,如净现值现值 mnmmnnnmijxxxxxxxxxxX212222111211)(njmixxymiijijij1112称矩阵称矩阵Y(yij)mn为向量归一标准化矩阵。矩为向量归一标准化矩阵。矩阵阵Y的列向量模等于的列向量模等于1,即,即 njymiij 1112注:向量注:向量归一标准化后归一标准化后 0yij1;正、逆向指标的方向没有发生变化。正、逆向指标的方向没有发生变化。0max1*ijmijxx令:令:mixxyjijij 1*对于负向指标对于负向指标fj,取:,取:ijmijxx 1*min令:令:
3、mixxyijijj 1*称矩阵称矩阵Y(yij)mn为线性比例标准化矩阵。为线性比例标准化矩阵。注:经线性比例变换注:经线性比例变换后后 0yij1;所有指所有指标均化为正向指标;标均化为正向指标;最优值为最优值为1。ijmijijmijxxxx 101*min,max对于负向指标对于负向指标fj,取:,取:ijmijijmijxxxx 101*max,min令:令:njmixxxxyjjjijij110*0称矩阵称矩阵Y(yij)mn为极差变换标准化矩阵。为极差变换标准化矩阵。注:经极差变换注:经极差变换后后 0yij1;所有指标均所有指标均化为正向指标;化为正向指标;最优值为最优值为1,
4、最劣值为,最劣值为0。njmisxxyjjijij11其中其中:为样本均方差为样本均方差为样本均值为样本均值 mijijjmiijjxxmsxmx121)(111称矩阵称矩阵Y(yij)mn为标准样本变换矩阵。为标准样本变换矩阵。注:经标准样本变换注:经标准样本变换后后标准化矩阵的样本均标准化矩阵的样本均值为值为0,方差为,方差为1。分值分值(fj)(ai)a1a2a3a4 550.52000018002.2775.42100020008.1535.61800027005.2955.52000015000.264)(ijxX njmixxymiijijij1112 3727.04811.046
5、08.05056.04394.05139.05217.06736.04147.05308.04882.04204.03727.02887.05990.04550.06591.05839.06708.04811.05069.05056.03662.04671.064)(ijyY 05.075.06667.025.05714.05.01114167.0000001115.05.06667.002857.064)(ijyY 5556.07143.09.09524.06667.088.07778.01117407.072.05556.04286.06923.08571.01117143.08182.0
6、9524.05556.08.064)(ijyYijmijijmijxxxx 101*min,max对于负向指标对于负向指标fj,取:,取:ijmijijmijxxxx 101*max,min令:令:njmixxxxyjjjijij111)(990*0变换变换后后 1yij100;所有指标均化为正向所有指标均化为正向指标;指标;最优值为最优值为100,最劣值为,最劣值为1。不重要时不重要时比比当当同样重要时同样重要时与与当当重要时重要时比比当当jijijiijffffffa05.01显然:显然:1,5.0 jiijiiaaa注意:注意:评分时应满足比较的传递性,即若评分时应满足比较的传递性,即若
7、f1比比f2重要,重要,f2又比又比f3重要,则重要,则f1比比f3重要。重要。),2,1(111niaawminjijnjiji fifif1f2f3f4f5f6wif1f2f3f4f5f6评分值评分值41.51.51.545.5:182/91/121/121/122/911/36 同样重要时同样重要时与与当当较为重要(或相反)时较为重要(或相反)时比比当当或或重要(或相反)时重要(或相反)时比比当当或或jijiiiiffffffr1)21(2)31(31并赋以并赋以rn=1。),2,1(1nikkwniiii fi比率值比率值wif1f2f3f4f5f611/21/61/61/61/25/
8、21/51/151/151/151/52/5 njmiyypmiijijij,2,1,2,11v计算第计算第j个指标的熵值,个指标的熵值,其中,其中,k0,ej0)1(ln1njppkemiijijj njegjj 11v确定指标权重。第确定指标权重。第j个指标的权重为个指标的权重为)1(1njggwnjjjj 5556.07143.09.09524.06667.088.07778.01117407.072.05556.04286.06923.08571.01117143.08182.09524.05556.08.064)(ijyY归一化处理得归一化处理得:1923.025.02639.025
9、32.0225.02588.02692.035.02932.02659.025.02118.01923.015.02030.02279.03375.02941.03462.025.02399.02532.01875.02353.064)(ijpP得得:6773.06726.06887.06924.06813.06895.0654321eeeeee)1(ln1njppemiijijj 差异系数:差异系数:3227.03274.03113.03076.03187.03105.0654321gggggg指标权重为:指标权重为:1651.01640.01679.01688.01661.01681.06
10、54321wwwwww)1(,21miwwwinii v对对m个专家给出的权重估计值平均,得到平个专家给出的权重估计值平均,得到平均估计值均估计值)1(11njwmwmiijj v计算估计值和平均估计值的偏差计算估计值和平均估计值的偏差)1,1(njmiwwijijij )1(njwj ,),(21TnwwwW njjw11)1(,1miywunjijji imiuau 1*max)(,)3.0,2.0,1.0,1.0,1.0,2.0(TW 737.0851.0707.0835.04321uuuu 21*211*1max,maximiimixxxx 212111min,minimiimixxx
11、x 则:则:理想解理想解为为 A*(x*1,x*2);负理想解负理想解为为 A-(x-1,x-2)。f1f2OA1A2A3AmA*A-问题:问题:如何表示各方案目标值靠近理想解和如何表示各方案目标值靠近理想解和远离负理想解的程度?远离负理想解的程度?定义方案定义方案ai与理想解、负理想解的相对贴近与理想解、负理想解的相对贴近度为度为 212212*jjijijjijixxSxxS和和),2,1(*miSSSCiiii 满足:满足:0Ci*1;理想点理想点Ci*1,负理想点负理想点Ci*0;方案逼近理想解而远离负理想解时方案逼近理想解而远离负理想解时Ci*1。nmijjnmijywvV *2*1
12、11,)|min(),|max(nijmiijmivvvJjvJjvV 理理想想解解 nijmiijmivvvJjvJjvV,)|max(),|min(2111负负理理想想解解正向指标集正向指标集负向指标集负向指标集 )1(1212*mivvSvvSnjjijinjjiji 和和),2,1(*miSSSCiiii 1118.00962.00461.00506.00439.01028.01565.01347.00415.00531.00488.00841.01118.00577.00599.00455.00659.01168.02013.00962.00507.00506.00366.00934
13、.0)(nmijjywV正正正正正正 负!负!正正正正,)3.0,2.0,1.0,1.0,1.0,2.0(TW 0458.0,0920.0,0439.0,0984.01009.0,0580.0,1197.0,0546.04321*4*3*2*1 SSSSSSSS负理想解负理想解理想解理想解1118.0,0577.0,0599.0,0455.0,0366.0,0841.02013.0,1347.0,0415.0,0531.0,0659.0,1168.0 VV312.0,614.0,268.0,643.0*4*3*2*1 CCCC,),(21TnwwwW 最优的权重最优的权重系数应满足:系数应满足
14、:),2,1(,01.min1112*njwwtsvvZjnjjminjjij符号含义与理符号含义与理想解法相同想解法相同 )(18.71112*1112*mijijnjmijijjyyyyw *2*111,)|min(),|max(nijmiijmiyyyJjyJjyY 理理想想解解),2,1()(122*miwyydnjjjiji 05.075.06667.025.05714.05.01114167.0000001115.05.06667.002857.0)(nmijyY正正正正正正负!正负!正正正 TW1236.01853.01534.02275.01461.00.1641,),2,1(
15、)(122*miwyydnjjjiji 得得0598.0,0616.0,1014.0,0553.04321 dddd功效系数为:功效系数为:)(hjx)(sjx6040)()()(sjhjsjijijxxxxd满意值的功效系数为满意值的功效系数为100,不允许值的功效系,不允许值的功效系数数60。njijjidwd1TnwwwW),(21,)3.0,2.0,1.0,1.0,1.0,2.0(TW 550.52000018002.2775.42100020008.1535.61800027005.2955.52000015000.264)(ijxX负!负!60809067.867086.82801
16、0010010067.766060606060100100100808067.866043.7164)(ijxX24.75,67.83,72,95.824321 dddd-4-2024-4-2024 mnmmnnnxxxxxxxxxXXXX21222211121121),(其中其中)1(),(21njxxxXTnjjjj )1(22111njXlXlXlXlZnnjjjntjtjj 满足满足)1(122221njlllnjjj ),1,(0),cov(njijiZZji )var()var()var(21nZZZ njjnjjXZ11)var()var(nkjjZ1)var(很小时,用很小时,
17、用Z1、Z2、Zk就就 可基本上反映出原始可基本上反映出原始n个指标所包含的信息量。个指标所包含的信息量。v优点:优点:减少了评价指标个数;减少了评价指标个数;充分保留了原始指标的信息量;充分保留了原始指标的信息量;新指标彼此不相关,避免了信息的新指标彼此不相关,避免了信息的交叉和重叠。交叉和重叠。其中其中:)1(),(21njlllLTnjjjj 数理统计已经证明,原始指标的第数理统计已经证明,原始指标的第j个主成分个主成分Zj为:为:)1(22111njXlXlXlXlZnnjjjntjtjj njmisxxyjjijij11其中其中:为样本均方差为样本均方差为样本均值为样本均值 miji
18、jjmiijjxxmsxmx121)(111 mtjtjitiijmttjtijjiiijijxxxxmyymr11)()(1111 其中:其中:R是对称矩阵,且主对角线元素均为是对称矩阵,且主对角线元素均为1,即:,即:),11njirrriijiij (,0|IR 解出解出n个特征值:个特征值:再由齐次线性方程组再由齐次线性方程组0)(LIR 解出对应的特征向量:解出对应的特征向量:)1(),(21njlllLTnjjjj 11)(nttjjb 并按累积贡献率准则,即以累积贡献率并按累积贡献率准则,即以累积贡献率%85)()(1111 nttkjjkjjb 为准则,提取为准则,提取k个主成
19、分个主成分)1(1kjYlZntjtjj kjjjZbZ1以以Z值的大小来评判被评价对象的优劣。值的大小来评判被评价对象的优劣。指标指标企业企业xi1xi2xi3xi4xi5xi6xi7xi8123456789101112131440.425.013.222.334.335.622.048.440.624.812.51.832.638.524.712.73.36.711.812.57.813.417.18.09.70.613.99.17.211.23.95.67.116.49.910.919.89.84.20.79.411.36.111.04.33.77.116.710.29.919.08.9
20、4.20.78.39.58.312.94.46.08.022.812.610.929.711.94.60.89.812.238.720.25.57.48.929.317.613.939.616.26.51.113.316.42.4423.5420.5780.7161.7263.0170.8471.7722.4490.7890.8740.0562.1261.32720.09.13.67.327.526.610.617.835.813.73.91.017.111.6 81141jisxxyjjijij)8,2,1()(131)8,2,1(1411412141 jxxsjxxijijjiijjyij
21、0.9574-0.2296-1.1391-0.43770.48720.5874-0.46081.57400.9728-0.2450-1.1930-2.01770.33310.81092.34880.3200-1.26920.69440.16790.2862-0.50840.43841.0639-0.4746-0.1872-1.72570.5229-0.2886-0.38120.4213-1.0433-0.7022-0.40131.46460.16050.36112.14670.1404-0.9831-1.68530.06020.4414-0.49430.4972-0.8585-0.9800-0
22、.29201.65060.33530.27462.11600.0723-0.8788-1.587-0.04910.1937-0.36950.2449-0.8904-0.6767-0.40961.56730.2048-0.02222.48890.1113-0.8637-1.3713-0.16920.1554-0.58270.5503-0.8980-0.7108-0.5631.44690.2942-0.07042.46170.1562-0.7994-1.3315-0.12950.17590.53721.8532-0.8845-0.7570.17581.3683-1.34010.21830.8436
23、-0.6896-0.6111-1.36660.5453-0.19270.5244-0.5512-1.0939-0.72881.26441.1756-0.40320.30732.0834-0.0973-1.0643-1.35040.2382-0.3045rij1 10.76120.76120.70760.70760.64280.64280.59640.59640.54430.54430.63120.63120.77290.77290.76120.76121 10.51490.51490.47540.47540.46660.46660.41950.41950.74080.74080.68020.6
24、8020.70760.70760.51490.51491 10.98790.98790.97770.97770.97410.97410.68420.68420.78020.78020.64280.64280.47540.47540.98790.98791 10.98070.98070.97980.97980.68810.68810.77310.77310.59640.59640.46660.46660.97770.97770.98070.98071 10.99240.99240.62650.62650.78700.78700.54430.54430.41950.41950.97410.9741
25、0.97980.97980.99230.99231 10.62900.62900.72450.72450.63120.63120.74080.74080.68420.68420.68810.68810.62650.62650.62900.62901 10.61960.61960.77290.77290.68020.68020.78020.78020.77310.77310.78700.78700.72450.72450.61960.61961 1)8,1(131141 jiyyrttjtiijj6.09121.01560.43320.21200.14200.01170.00300.0013L1
26、j0.32370.39790.4596-0.66200.1174-0.11800.1501-0.1930L2j0.28390.6214-0.10700.2812-0.65200.1191-0.05000.0296L3j0.3905-0.22400.0230-0.2390-0.0990-0.1050-0.55300.6426L4j0.3856-0.2730-0.0530-0.10100.01500.85700.1268-0.1130L5j0.3804-0.31200.02040.1295-0.1830-0.32500.73230.2564L6j0.3716-0.3630-0.09500.0459
27、-0.2160-0.3240-0.3310-0.6860L7j0.32210.2883-0.7470-0.07000.4812-0.12300.05240.0042L8j0.35630.13560.45360.62570.4896-0.0030-0.13300.0077bj0.76140.13820.05410.02650.01780.00150.00040.0002特征值:特征值:85.08996.01382.07614.021 bb提取第提取第1 主 成主 成分:分:8765432188122111113563.03221.03716.03804.03856.03905.02839.032
28、37.0YYYYYYYYYlYlYlZ 第第 2 主主成分:成分:8765432188222211221356.02883.03630.03120.02730.02240.06214.03979.0YYYYYYYYYlYlYlZ -2.8488-2.8488企业企业1234567Z10.73561.0895-2.8488-2.04770.06453.4825-0.2922Z22.69920.0774-0.6040-0.05721.0097-0.8192-1.0618Z按按Z1排序排序按按Z排序排序企业企业891011121314Z11.01315.1459-0.3182-2.3927-4.39
29、440.40220.3607Z20.8683-1.1556-0.77430.0824-0.76930.7407-0.2363Z按按Z1排序排序按按Z排序排序1 12 23 34 45 56 67 78 89 9101011111212131314142211ZbZbZ 0.93310.93310.84020.8402-2.2525-2.2525-1.5670-1.5670 0.18860.18862.53832.5383-0.3692-0.36920.89140.89143.75943.7594-0.3493-0.3493-1.8104-1.8104-3.4522-3.4522 0.40860
30、.40860.24200.24201 12 23 34 45 56 67 78 89 910101111121213131414 43212211RRRRvcvcvcNRnn称为称为n维物元,简记为维物元,简记为R其中:其中:TnTnvvvVcccC),(,),(2121 ),()(uKA)(uKA)(),(:)(uKuUuKAA称为可拓子集称为可拓子集 的关联函数,简记为的关联函数,简记为 K(u)。的的可拓域可拓域;A的的非域非域。A)(21)(21),(00abbaxXxp 点与区间的距对于开区间、半开半闭区间同点与区间的距对于开区间、半开半闭区间同样适用样适用。),(),(),()(0
31、0XxpXxpXxpxK 则则 xX0 的充要条件是的充要条件是:K(x)0;xXX0 的充要条件是的充要条件是:-1K(x)0;x X 的充要条件是的充要条件是:K(x)-1。nnXcXcXcNVCNR002201100000),(其中其中N0 表示标准产品,表示标准产品,ci(i=1,2,n)表示产表示产品评价指标,品评价指标,X0i=a0i,b0i(i=1,2,n)表示标准产品评价指表示标准产品评价指标的经典域。标的经典域。nnXcXcXcNVCNR 2211),(其中其中N 表示节域产品,即包括标准产品和可拓表示节域产品,即包括标准产品和可拓性产品。可拓性产品是指能转化为标准产品的性产
32、品。可拓性产品是指能转化为标准产品的产品。产品。Xi=ai,bi(i=1,2,n)表示产品评价指标表示产品评价指标的节域。的节域。nnBBBBxcxcxcNVCNR2211),(其中其中NB表示待评产品,表示待评产品,xi表示待评产品关于指表示待评产品关于指标标ci(i=1,2,n)的指标值。的指标值。为各指标的权系数。为各指标的权系数。,综合关联函数综合关联函数iniiiiiiiiidxKdxKniaaaxxK 100)()(),2,1(,)(10,5.710,5.610,610,7,10,810,710,710,84321432100ccccNRccccNR现有二产品现有二产品,其待评物元矩阵分别是:其待评物元矩阵分别是:9768,75.65.8943214321ccccNRccccNRBBAAa0iaixi计算得:计算得:2010)()()()(,215.11)()()()(43214321xKxKxKxKxKxKxKxKBBBBAAAA)4,3,2,1(,)(00 iaaaxxKiiiiii 取评价指标的权重系数分别为:取评价指标的权重系数分别为:d1=0.4,d2=0.2,d3=0.1,d4=0.3得:得:KA=0,KB=0.4,均符合标准产品要求。,均符合标准产品要求。