You are on page 1of 18

Data Envelopment Analysis

Weights Optimization
( )
0
0
1
1 1
1
.
0 1,...,
2
1
1,...,
1,...,
s
r rJ
r
s m
r rj i ij
r i
m
i iJ
i
r
i
Max u y
s t
u y v x j N
M
v x
u r s
v i m
c
c
=
= =
=

`
)
s =
=
> =
> =

Primal Dual Relations


( )
0
0
1
1 1
1
.
0 1,...,
2
1
1,...,
1,...,
s
r rJ
r
s m
r rj i ij
r i
m
i iJ
i
r
i
Max u y
s t
u y v x j N
M
v x
u r s
v i m
c
c
=
= =
=

`
)
s =
=
> =
> =

( )
0
0
0
1 1
0
1
1
0
.
0 1,...,
3 1,...,
0 1,...,
, 0 1,..., ; 1,...,
free
m s
i r
i r
N
j ij iJ i
j
N
j rj r rJ
j
j
i r
Min k S S
s t
x k x S i m
M y S y r s
j N
S S i m r s
k
c

+
= =

=
+
=
+

(
+
`
(

)
+ = =
= =
> =
> = =

Minimize Inputs Interpretation


( )
0
0
0
1 1
0
1
1
0
.
0 1,...,
3 1,...,
0 1,...,
, 0 1,..., ; 1,...,
free
m s
i r
i r
N
j ij iJ i
j
N
j rj r rJ
j
j
i r
Min k S S
s t
x k x S i m
M y S y r s
j N
S S i m r s
k
c

+
= =

=
+
=
+

(
+
`
(

)
+ = =
= =
> =
> = =

Graphic Illustration

Conclusions From Optimality
* *
0
i r
S S

+
= =
*
0
1 k =
Pareto Efficient
*
0
k
Technical Input Efficiency
*
1
*
1
1,...,
1,...,
N
t
i j ij
j
N
t
r j rj
j
x x i m
y y r s

=
=
= =
= =

Efficient Goal and


Efficient Peers
Maximize Outputs
( )
0
0
0 0
0
1 1
1
1
.
1,...,
4 0 1,...,
0 1,...,
, 0 1,..., ; 1,...,
free
m s
J i r
i r
N
j ij i iJ
j
N
j rj J rJ r
j
j
i r
J
Max h I O
s t
x I x i m
M y h y O r s
j N
I O i m r s
h
c
o
o
o
= =
=
=

(
+ +
`
(

)
+ = =
= =
> =
> = =

Conclusions From Optimality


0
*
1
J
h =
* *
0
r i
O I = =
Pareto Efficient
0
*
1
J
h
Technical Output Efficiency
0
*
*
0
1
J
k
h
=
Return to Scale
Modeling Return to Scale
( )
0
0
0
1 1
0
1
1
1
0
.
0 1,...,
5 1,...,
1
0 1,...
, 0 1,..., ; 1,...,
free
m s
i r
i r
N
j ij iJ i
j
N
j rj r rJ
j
N
j
j
j
i r
Min k S S
s t
x k x S i m
M y S y r s
j N
S S i m r s
k
c

+
= =

=
+
=
=
+

(
+
`
(

)
+ = =
= =
=
> =
> = =

( )
0
0
0
1 1
0
1
1
0
.
0 1,...,
3 1,...,
0 1,...,
, 0 1,..., ; 1,...,
free
m s
i r
i r
N
j ij iJ i
j
N
j rj r rJ
j
j
i r
Min k S S
s t
x k x S i m
M y S y r s
j N
S S i m r s
k
c

+
= =

=
+
=
+

(
+
`
(

)
+ = =
= =
> =
> = =

Numeric Example
Numeric Example
( )
{ }
1 2
1 2 1 2 3 4
1 2 1 2 3 4
1 2 1 2 3 4
1 2 1 2 3 4
1 2 3 4
1 2
1 2 3 4
3 10
.
3 10 370 13 14 5 0
5 40 320 12 16 3 0
7 2 30 410 19 13 4 0
3 50 190 10 15 2 0
370 13 14 5 1
,
, , ,
Max u u
s t
u u v v v v
u u v v v v
M u u v v v v
u u v v v v
v v v v
u u
v v v v
c
c
+
+ s
+ s
+ s
+ s
+ + + =
>
>
Numeric Example
( )
{ }
1 2
1 2 1 2 3 4
1 2 1 2 3 4
1 2 1 2 3 4
1 2 1 2 3 4
1 2 3 4
1 2
1 2 3 4
3 10
.
3 10 370 13 14 5 0
5 40 320 12 16 3 0
7 2 30 410 19 13 4 0
3 50 190 10 15 2 0
370 13 14 5 1
,
, , ,
Max u u
s t
u u v v v v
u u v v v v
M u u v v v v
u u v v v v
v v v v
u u
v v v v
c
c
+
+ s
+ s
+ s
+ s
+ + + =
>
>
1
2
1
1 2
2
3
4
0.199
0.001
0.001
3 10 0.607
0.001
0.044
0.001
u
u
v
u u
v
v
v
| | | |
| |
| |
| |
= + =
| |
| |
| |
| |
| |
\ . \ .
Numeric Example
( )
{ }
1 2
1 2 1 2 3 4
1 2 1 2 3 4
1 2 1 2 3 4
1 2 1 2 3 4
1 2 3 4
1 2
1 2 3 4
5 40
.
3 10 370 13 14 5 0
5 40 320 12 16 3 0
8 2 30 410 19 13 4 0
3 50 190 10 15 2 0
320 12 16 3 1
,
, ,
Max u u
s t
u u v v v v
u u v v v v
M u u v v v v
u u v v v v
v v v v
u u
v v v v
c
c
+
+ s
+ s
+ s
+ s
+ + + =
>
>
1
2
1
1 2
2
3
4
0.001
0.020
0.005
5 40 1
0.001
0.001
0.001
u
u
v
u u
v
v
v
| | | |
| |
| |
| |
= + =
| |
| |
| |
| |
| |
\ . \ .
Numeric Example
( )
{ }
0 1 2 3 4 1 2
1 2 3 4 0 1
1 2 3 4 0 2
1 2 3 4 0 3
1 2 3 4 0 4
1 2 3 4 1
1 2 3 4
.
370 320 410 190 410 0
13 12 19 10 19 0
14 16 13 15 13 0
9 5 3 4 2 4 0
3 5 2 3 2
10 40 30 50
Min k S S S S S S
s t
k S
k S
k S
M k S
S
c






+ +

+
( + + + + +

+ + + + =
+ + + + =
+ + + + =
+ + + + =
+ + + =
+ + +
2
1 2 3 4
1 2 3 4 1 2
0
30
, , , 0
, , , , , 0
S
S S S S S S
k free

+
+ +
=
>
>
M
3
is inefficient
M
2
and M
4
are efficient
peers
DEA Software

Summary, Pros and Cons
Avoids arbitrary selection of weights.
Uses the theory of mathematical
programming.
Need to solve n programs.
Assumes deterministic measurements.
Requires a sufficient number of units.

Summary, Pros and Cons
Requires units to use the same exact
variables.
Insufficient conclusions for decision
making.
Use of discrete or even categorical values.
Unchangeable values.

You might also like