In: Operations Management
The probabilities come from the historical data a company may have. For example, if the states of nature are high, moderate, or low demands, the company will check out of the last 10 to 20 years, how many times a high demand, and moderate demand, or low demand has occurred. Based on this frequency distribution, the probabilities will be computed. This will be called the prior probability (probability based on prior knowledge). Say, the historical data exists for the last 20 years and the occurrence of high demand was only for five years out of these 20 years. So, the probability of high demand will be 5/20 = 0.25
Sometimes, things may not be so simple. The prior estimates of probabilities may be inaccurate for various reasons. The company may not have adequate data to compute the prior probabilities accurately. This can happen for a new product introduction. The demand for this new product may be very different from the other products that have been introduced by the company in the past. Secondly, the nature of the economy and market may have changed in a way that prior estimates are not even valid. In all these cases, the company would like to conduct a research (generally by the third party) and use posterior probabilities which will be the outcome of the research in place of the prior probabilities.