Ⅳ．TOPSIS 法
1. TOPSIS 法所用程序
所用程序文件为 9MethodsHos-4.sps：
*Hospital evaluation by 9 methods; Filename: 9MethodsHos-4.sps.
*--------------------------------------------------------------------------.
* Method-4: TOPSIS:.
GET FILE='c:spsssav9methodsHos.sav'.
*(1) The variables are made into the same trend (high good); So,
the following "i=100-i" is high good:.
COMPUTE i=100-i.
* Calculating the square value for each variable:.
COMPUTE a2=a**2.
COMPUTE b2=b**2.
COMPUTE c2=c**2.
COMPUTE d2=d**2.
COMPUTE e2=e**2.
COMPUTE f2=f**2.
COMPUTE g2=g**2.
COMPUTE h2=h**2.
COMPUTE i2=i**2.
FREQUENCIES VARIABLES=a2 b2 c2 d2 e2 f2 g2 h2 i2
/FORMAT=NOTABLE /STATISTICS=Sum.
*-----------------------------------------------------------.
*(2) Calculating the "standardized value" for each variable:.

COMPUTE za=a/SQRT(29196.27).
COMPUTE zb=b/SQRT(49007.50).
COMPUTE zc=c/SQRT(32747.97).
COMPUTE zd=d/SQRT(28520.21).
COMPUTE ze=e/SQRT(39665.69).
COMPUTE zf=f/SQRT(48629.00).
COMPUTE zg=g/SQRT(40497.02).
COMPUTE zh=h/SQRT(42645.44).
COMPUTE zi=i/SQRT(42794.36).
*----------------------------------------------------.
*(3) Finding best point (Zplus) & wost point (Zminus).
*Ranking: When high-good: (A); Wost to best.

SORT CASES BY za(A).
LIST VARIABLES=hos za.
SORT CASES BY zb(A).
LIST VARIABLES=hos zb.
SORT CASES BY zc(A).
LIST VARIABLES=hos zc.
SORT CASES BY zd(A).
LIST VARIABLES=hos zd.
SORT CASES BY ze(A).
LIST VARIABLES=hos ze.
SORT CASES BY zf(A).
LIST VARIABLES=hos zf.
SORT CASES BY zg(A).
LIST VARIABLES=hos zg.
SORT CASES BY zh(A).
LIST VARIABLES=hos zh.
SORT CASES BY zi(A).
LIST VARIABLES=hos zi.
*-----------------------------------------------------------.
*(4) Calculating the "Euclid distance" (Dplus & Dminus).
*Dplus: The lowest distance (to the best) is the best;.
*Dminus: The largest distance (to the wost) is also the best.

SORT CASES BY hos(A).
COMPUTE Dplus=SQRT((za-0.37)**2+(zb-0.44)**2+
(zc-0.41)**2+(zd-0.38)**2+
(ze-0.42)**2+(zf-0.43)**2+
(zg-0.37)**2+(zh-0.43)**2+
(zi-0.44)**2).
FORMAT Dplus(F7.2).
LIST VARIABLES=hos Dplus /FORMAT=NUMBERED /CASES FROM 1 TO 5.

COMPUTE Dminus=SQRT((za-0.50)**2+(zb-0.45)**2+
(zc-0.48)**2+(zd-0.51)**2+
(ze-0.48)**2+(zf-0.45)**2+
(zg-0.48)**2+(zh-0.47)**2+
(zi-0.46)**2).
FORMAT Dminus(F7.2).
LIST VARIABLES=hos Dminus /FORMAT=NUMBERED /CASES FROM 1 TO 5.
*----------------------------------------------.
*(5) Calculating Ci: Highest is the best point:.

COMPUTE Ci=Dplus/(Dplus + Dminus).

RANK VARIABLES=Ci(D) /RANK INTO No1.
SORT CASES BY No1(A).
FORMATS Ci(F7.2) No1(F3.0).
LIST VARIABLES=hos Ci No1 /FORMAT=NUMBERED /CASES=FROM 1 TO 5.
*----------------------------------------------------------------------------.
2. TOPSIS 法的计算方法与步骤
与上法类似。计算各评价单元指标值与最优值的相对接近程度。
Ci 值 （输出为 CI 值）的计算公式为 （见程序文件的第（5）部分）：
Ci = Dplus /(Dplus + Dminus) （i = 1,2,…,m）
TOPSIS 法按接近度大小对各评价单元优劣进行排序，Ci值越大，表明其越接近最优水平。接近度取值在 0 与 1 之间，该值越接近 1 ，说明所评价单元接近最优水平的程度愈高，反之，该值越接近 0 ，说明所评价单元接近最优水平的程度愈低。
3. TOPSIS 法的评价结果
算得5 家医院的排序由好到差依次为T—R—S—Q—P，Ci 值（输出为 CI值）见表 5-2 。
---------------------------------------------
HOS CI NO1
---------------------------------------------
1 T 0.86 1
2 R 0.83 2
3 S 0.67 3
4 Q 0.31 4
5 P 0.09 5
---------------------------------------------