Uncertainty and Stochastic Programming Sample Clauses

Uncertainty and Stochastic Programming. ▇▇▇▇▇▇▇ (1994) defines that randomness can be replaced with the expression of uncertainty. It can be described as lack of predictability of what will happen. Randomness is divided into two categories: external and internal randomness. External randomness refers to randomness that we cannot control. An example could be the probability of an earthquake within 5 years. Internal randomness refers to ignorance, to our lack of knowledge. An example can be the probability that France had a net export of goods to Germany last year. Estimation can be done in two ways: distributional and singular. In distributional mode we try to understand a random event by analyzing the cases which are similar and occurred previously, in singular mode we try to understand the event by analyzing it directly. By definition, stochastic is a problem in which the data and parameters are not known with certainty, but a probability distribution is known (▇▇▇▇ and ▇▇▇▇▇▇▇, 1989). In other words, stochastic programming allows us taking the uncertainty into consideration. Since there are many challenges and unpredictable situations in the industries and managers need to make decisions under uncertainty, stochastic programming models are used in variety of applications. We need to underline the fact that stochastic problems are one of the most complicated optimization problems. Usually models are firstly built as deterministic models and then turned to stochastic models when the decision-maker realizes the shortcomings of the model when representing the real system. Reconstructing the deterministic model to stochastic model implies redefinition of the objective function as well. (▇▇▇▇▇▇▇, 1994) Scenario trees are often very important in decision analysis and stochastic programming. It consists of nodes and each node in the tree represents state of the world at a particular point in time. Decisions are made at these nodes. The scenario tree branches off for each possible value of a random variable in each period. (▇▇▇▇▇▇▇ and ▇▇▇▇▇▇▇, 2001)