Decision Theory
Lecture 1:
Introduction
Decisions
· should be made formally/systemetically
Normative or
prescriptive
->
Identifying the best decision to take
steps ofproblem
"d
sorving
1. dentify and define the problem
& 2. Determine of
alternative solutions
set
will be used to evaluate
3. Determine critere that alternatives
5.
4. Evaluate the alternatives
⑳e choose an alternative
. 6. Implementserected
7. Evaluate results if
alternative
satisfactory
Positive or descriptive
->
describes people
what actually do
Homo Economicus /Economic Man
·
emotionless individual, only self-interested, attempts maximizing expected
Utility
unlimited
·
cognitive abilities & unlimited willpower
·
rational a consistent
Rational always right. Irrational
right.
->
decisions are not decisions can be
-> Because of
unknown unknowns
Model Development
iconic modes physical replicas
*
analog moders- don'thave same physical appearance
dex. Thermometer representing temperature
mathematical moder
* representation ofproblem bysystem of
symbols &
mathematical relationships or expressions
Advantage ofModels:requires less time, less expensive, less risk
, Mathematical Models
expression thatdescribes problem's objective objective function
· -
limited time available requires capcity constraints
· =
ex.
complete mathematical model:
Maximize p=10x objective function
subject
5,40 Iconstraints
to
->
deterministic model if uncontrollable inputs are unknown d cannot vary
->
stochastic or probabilistic if inputs are uncertain or
subjet to variation
Linear programming
Linear: Non linear:
1
er 1 +
4. I
4 y2
+
Lecture 2:
-
Largrange
Largrange function.
Step 1:Write the
Start constraints like followed.
with
objective function and add
L 0.20x
0.25, +v(+4-50001 ply+s40001 4+ s3000
=
9
+ + +
3
4 y +
5000
=
-
5000
- 4000
-
-
4000 set all constraints equal to zero
↑ 3000 3000
-
· with every
constraintadd another variable
=
·
if
=
the constraint equal but, or one
is not has to add a
S to each constraint
Step 2:Mahe partial derivatives ofall variables.
Lecture 3:
Influence Diagrams
Rectangle or
Squares decision nodes
Circal or ovals chance nodes
Diamonds consequence nodes
Lecture 1:
Introduction
Decisions
· should be made formally/systemetically
Normative or
prescriptive
->
Identifying the best decision to take
steps ofproblem
"d
sorving
1. dentify and define the problem
& 2. Determine of
alternative solutions
set
will be used to evaluate
3. Determine critere that alternatives
5.
4. Evaluate the alternatives
⑳e choose an alternative
. 6. Implementserected
7. Evaluate results if
alternative
satisfactory
Positive or descriptive
->
describes people
what actually do
Homo Economicus /Economic Man
·
emotionless individual, only self-interested, attempts maximizing expected
Utility
unlimited
·
cognitive abilities & unlimited willpower
·
rational a consistent
Rational always right. Irrational
right.
->
decisions are not decisions can be
-> Because of
unknown unknowns
Model Development
iconic modes physical replicas
*
analog moders- don'thave same physical appearance
dex. Thermometer representing temperature
mathematical moder
* representation ofproblem bysystem of
symbols &
mathematical relationships or expressions
Advantage ofModels:requires less time, less expensive, less risk
, Mathematical Models
expression thatdescribes problem's objective objective function
· -
limited time available requires capcity constraints
· =
ex.
complete mathematical model:
Maximize p=10x objective function
subject
5,40 Iconstraints
to
->
deterministic model if uncontrollable inputs are unknown d cannot vary
->
stochastic or probabilistic if inputs are uncertain or
subjet to variation
Linear programming
Linear: Non linear:
1
er 1 +
4. I
4 y2
+
Lecture 2:
-
Largrange
Largrange function.
Step 1:Write the
Start constraints like followed.
with
objective function and add
L 0.20x
0.25, +v(+4-50001 ply+s40001 4+ s3000
=
9
+ + +
3
4 y +
5000
=
-
5000
- 4000
-
-
4000 set all constraints equal to zero
↑ 3000 3000
-
· with every
constraintadd another variable
=
·
if
=
the constraint equal but, or one
is not has to add a
S to each constraint
Step 2:Mahe partial derivatives ofall variables.
Lecture 3:
Influence Diagrams
Rectangle or
Squares decision nodes
Circal or ovals chance nodes
Diamonds consequence nodes
