RISK ASSESSMENT DEPT. OF AGROINDUSTRIAL TECHNOLOGY FACULTY OF AGRICULTURAL TECHNOLOGY UNIVERSITAS BRAWIJAYA

Size: px

Start display at page:

Download "RISK ASSESSMENT DEPT. OF AGROINDUSTRIAL TECHNOLOGY FACULTY OF AGRICULTURAL TECHNOLOGY UNIVERSITAS BRAWIJAYA"

Jayson Turner
5 years ago
Views:

1 RISK ASSESSMENT DEPT. OF AGROINDUSTRIAL TECHNOLOGY FACULTY OF AGRICULTURAL TECHNOLOGY UNIVERSITAS BRAWIJAYA

2 RISK ASSESSMENT To determine relative priority and get information to solve it Risk that should be assessed/measured: Dimension (frequency, number of lossess) Determine how serious if the risk happens be carefull with the effect of risk Risk assessment by probability distribution EOL, EVPI, EPV

3 RISK ASSESSMENT METHODS 1. EOL (Expected Opportunity Loss) A number of loss pay off because we dont choose the alternative action priciple: minimize losses because we choose a specific alternative Rationally, we choose the smallest EOL value to avoid the dissatisfaction/regret in the future

4 EXAMPLE OF EOL An entrepreneur wanna invest his money to one of this business alternatives: shrimp ponds, food stalls and music studios, with consideration of good economic growth rate (probability 0.4) and bad (0.6). Data seen in Table, what is the best decision for the entrepreneur based on EOL????

5 FIRST TABLE OF EXPECTED OPPORTUNITY LOSS Business Alternative Economic Growth rate Good (0,4) Bad (0,6) Shrimp ponds Food stalls Music studio

6 FIRST TABLE OF EXPECTED OPPORTUNITY LOSS Business Alternative Economic Growth rate Good (0,4) Bad (0,6) Shrimp ponds Food stalls Music studio

7 PAY OFF TABLE (COMPARED WITH EVERY BUSINESS ALTERNATIVES) BIGGEST VALUE: OPPORTUNITY LOSS Business Alternative Economic growth rate Good (0,4) Bad (0,6) Shrimp ponds Food stalls Music studio

8 EOL CALCULATION EOL (Shrimp ponds) = (0,4) (0,6) = EOL (food stalls) = (0,4) + 0(0,6) = EOL (music studio) = 0(0,4) (0,6) = *** The chosen business with the smallest opportunity loss is Music Studio with EOL min = 6.000

9 CHOOSE THE BEST BUSINESS ALTERNATIVES WITH EOL? Business alternatives Apples beverages Apple Jenang Economic conditions quiet Normal crowd (0,10) (0,7) (0,2) Apple chips

10 PAY OFF TABLE Business alternatives quiet (0,10) Economic conditions normal (0,70) crowd (0,2) Apples beverages Apple Jenang Apple chips

11 EOL (AB) = (0,1) (0,7) (0,2) = EOL (AJ) = (0,1) (0,2) = ** (CHOOSEN) EOL (AC) = 0 (O,1) (0,7)+0 (0,2) =

12 EVPI METHOD (EXPECTED VALUE PERFECT INFORMATION) It is a difference between the perfect information and the minimum risks (EOL) As the maximum amount paid for obtaining the perfect information (equal to the minimum EOL) EVPI = EV (Perfect) - EV (min risk) The decision determines the EV (Perfect) depending on the type of actor: RISK TAKER (max) and RISK AVOIDER (min)

13 EXAMPLE OF EVPI Business alt Economic growth Good (0,4) Bad (0,6) Shrimp ponds Food stalls Music Studio** Based on choosen business alternative (music studio), the EVPI = EV(perfect) biggest value from the first table EV(PERFECT) = (0,4) (0,6) = = EV(min) choosed based on EOL min (in this case = music studio) EV (min) = (0,4) (0,6) = = EVPI = EV(S) EV(min) = EOL min = = 6.000

14 Determine the value of EVPI based on business alternatives (apple beverage, apple jenang and apple chips) selected...????? Condition business quiet Normal crowd (0,10) (0,7) (0,2) Apple beverages Apple jenang Apple chips

EVPI EV (perfect) =125.000(0,1)+440.000(0,7)+750.000(0,2) = 12.500+308.000+150.000 =470.

15 EVPI EV (perfect) = (0,1) (0,7) (0,2) = = EV (choosen based on EOL min) = (0,1) (0,7) (0,2) = EVPI = EV (perfect) EV(min) = = = EOL min

16 UNCERTAINTY THEORY The probability of a result is not known for certain (only by estimate) Decision-makers do not have complete information The thing to be decided has never happened Anticipate conditions of uncertainty: 1. looking for more information 2. through research 3. the use of subjective probabilities

17 DECISION-MAKING METHOD OF UNCERTAINTY CONDITIONS There are 5 criterias: 1. Laplace (Bayesian) 2. Maximin Wald 3. Maximax 4. Hurwics 5. Minimax Savage/ Regret

18 LAPLACE (BAYESIAN) The approach gives the SAME value for each event/condition Similar with EQUAL LIKELIHOOD criteria Assumptions: the decision maker determines the same probability of each occurrence/condition Selected result: HIGHEST weight

19 LAPLACE Business alt Economic growth Good (0,5) Bad (0,5) Result Shrimp ponds (0,5) (0.5) = Food stalls (0,5) (0,5) = Music Studio** (0.5) (0.5) = ***

20 MAXIMIN WALD A pessimistic view (RISK AVOIDER) that avoids the risk of future results The hope is the CREDIT VALUE for the alternative chosen actions How to: compare MINIMUM and MAXIMUM Payoffs to get the most MINIMUM value selected

21 MAXIMIN WALD Business alt Economic growth result good bad Min Max Shrimp ponds Food stalls Music Studio** ***

22 MAXIMAX Contrary to Maximin's criteria Optimistic view (RISK TAKER), aggressive attitude with the greatest results to be achieved MAXIMIN and MAXIMAX Criteria is unrealistic for the real business world because of the PARSIAL assessment

23 MAXIMAX Business alt Economic growth Result good Bad Max Max Shrimp ponds Food stalls Music Studio** ***

24 HURWICS Criteria result of Maximin and Maximax compromise Rarely as Risk Taker and Avoider are perfect The coefficient of Optimism (α) and pessimism (1-α) Pay off weighted = α (pay off maximum) + (1- α) (pay off minimum) Best alternative = BIGGEST value

25 HURWICS Economic growth Business alt Maximax Maximin Result (0,6) (0,4) Shrimp ponds Food stalls Music Studio** (0,6) (0,4) = = (0,6) (0,4) = = (0,6) (0,4) = =80.000*

26 MINIMAX REGRET Based on Opportunity Loss Incidental Current/UNFORTUNATE EVENT CRITERION (comparing between income with specific action and highest expected income)

27 MINIMAX REGRET Business alt Economic growth Opp.Loss (between each business alt) good bad good Bad Result (max -> min) Shrimp ponds Food stalls Music Studio** *

28 THANK YOU

MULTICRITERIA DECISION AIDING

MULTICRITERIA DECISION AIDING ANTOINE ROLLAND, Université LYON II CERRAL, 24 feb. 2014 PLAN 1 Introduction 2 MCDA Framework 3 utility functions 4 Outranking approach 5 Other methods A. Rolland MCDA 2 /