lingo处理DEA的最基本的CCR模型，不是一个理想的选择，我建议还是用DEAP比较好。至于CCR的对偶模型，跟CCR模型有什么本质的区别吗？

CCR对偶模型可以得出系数值lamda在最优下的和，可以判断规模效率。DEAP我有尝试，但不知道dta为何物，如何数据输入。。。所以作罢。能指点一二么

关注。。。。。。。。。。。。。。

MODEL: ! Data Envelope Analysis of Decision Maker Efficiency ;
SETS:
  DMU/1..6/: ! Six schools;
     SCORE; ! Each decision making unit has a;! score to be computed;
  FACTOR/COST RICH WRIT SCIN/;! There is a set of factors, input & output;
  DXF( DMU, FACTOR):  F; ! F( I, J) = Jth factor of DMU I;
ENDSETS
DATA: ! Inputs are spending/pupil, % not low income;
! Outputs are Writing score and Science score;
  NINPUTS = 2;  ! The first NINPUTS factors are inputs;
!      The inputs,    the outputs;
F= 89.39        64.31        25.2        223
86.25        99        28.2        287
108.13        99.6        29.4        317
106.38        96        26.4        291
62.4        96.2        27.2        295
47.19        79.9        25.2        222;

ENDDATA
!----------------------------------------------------------;
! The Model;
SETS:
   ! Weights used to compute DMU I's score;
   DXFXD(DMU,FACTOR) : W;
ENDSETS
! Try to make everyone's score as high as possible;
  MAX = @SUM( DMU: SCORE);
! The LP for each DMU to get its score;
  @FOR( DMU( I):
   SCORE( I) = @SUM( FACTOR(J)|J #GT# NINPUTS:
    F(I, J)* W(I, J));
! Sum of inputs(denominator) = 1;
   @SUM( FACTOR( J)| J #LE# NINPUTS:
    F( I, J)* W( I, J)) = 1;
! Using DMU I's weights, no DMU can score better than 1;
   @FOR( DMU( K):
    @SUM( FACTOR( J)| J #GT# NINPUTS: F( K, J) * W( I, J))
     <= @SUM( FACTOR( J)| J #LE# NINPUTS:
      F( K, J) * W( I, J))
   )
  );
! The weights must be greater than zero;
  @FOR( DXFXD( I, J): @BND( .00001, X, 100000));
END

5# deman007
谢啦，不过这好像是元模型不是对偶模型的代码诶。。。您还知道对偶的代码么？