The Utility Value of Information in Pig Production

The possibilities for registrations and the corresponding costs are steadily increasing. We need to treat registrations and the use of Decision Support Systems (DSS) as a production factor in line with feed. The quality or utility value of a registration can be measured on the improvement in decisions. In order to do this we need, however, to define and categorize the decisions made in pig production. In the paper the evaluation of production control, registrations used in culling decisions, pregnancy test and weighing of slaughter pigs is presented. These informations have only a low value, but the analysis indicates how their use might be improved. The continuation of the efforts of utility evaluation is important, if we are not to overtax the producers with information. Introduction Traditionally several traits are registered in pig herds. In sow herds such traits comprises, e.g., litter size and event dates (mating, farrowing, etc.). It is generally accepted that these traditional traits are useful for decision purposes. In contrast few traits are measured in slaughter pig production. The reason for this difference in registration detail is not clear. With the advent of electronic equipment a whole new range of registrations becomes possible. The possibilities comprises electronic identification; automatic weighing; temperatureand activity measurements; and several registrations via video recordings using image analysis techniques. (Van der Stuyft et al., 1991). The general attitude towards these new registrations is that they will improve, either income of the pig producer; welfare of the pig; reduce the environmental impact of pig production; or help to fulfil consumer demands for quality certification. As with the traditional registrations, the possible benefits of these registrations are not directly estimated. It seems that the choice of registrations has a large random element. In the author’s opinion it is important to treat information as a production factor in line with, e.g., feed. We need to define the quality and value of information, just as we define quality and value of feed stuffs. The value of feed stuffs is measured by their effect on the output, i.e., daily gain, feed conversion and meat quality. The value of information should be measured similarly on the output, i.e., the improvement in decisions. The value of information is dependent on it’s applicability in the decision process, that is to say, to what extent the information helps the producers to reach their overall goal. Each decision in the herd has its own demand for information. The value of information thus depends on the decision context, like the value of feed stuffs differs, whether they are used for sows or slaughter pigs. Statistical decision theory using bayesian techniques gives the necessary theoretical tools for measuring improvement in decisions. However, a large effort is needed in defining and categorizing the decisions made in pig production, and in estimating the necessary probability distribution of the relevant traits. The purpose of this paper is to present Danish efforts in the field of information evaluation. The presented examples cover several aspects of pig production. General approach A part of the statistical decision theory originates in the so-called game theory describing the situation, where 2 gamblers can choose between several actions. The loss of a gambler ( and the gain of his opponent) both depends on his choice of actions and the action his opponent chooses. A pig producer Danish Informatics Network in the Agricultural Sciences