The two possible results of the ninth stage are: 1. A dull prognosis. 2. A pithy prognosis.
GOAL OF THE sTAgE:In the first case it is necessary to solve the problem whether to ignore that fact when the prognosis is unclear or to transfer to the first stage and to change the components in the technological chain.
In the second case it is necessary to carry out an objective information synthesis of a non-prognostic character obtained in this and other prognoses with the purpose of transforming it into the shape opportune for making decisions.
DESCRIPTION OF THE STAGE:
The first result means that a dull prognosis is obtained.
It is necessary to ignore it because the knowledge got from a such prognosis is disproportionately insignificant in comparison with the immense work made for getting it and to transfer again to the first stage and to change components in the technological chain.
The second result means that a pithy prognosis is obtained.
In the 10-th stage it is necessary to realize an objective information synthesis of a non-prognostic character and an information obtained in this one and other prognosis with the purpose of transferring it into a form suitable for making decisions. So, in case we are working up a long-term prognosis, middle-term and even short-term prognoses could be useful. The use of the entire totality of data enhances the quality of the decision made on the basis of the prognosis.
NOTE:
The use of the prognosis for making decisions is described in detail in the literature on the decision making. Among the Russian works, it is necessary to mention the research works of O.I. Larichev [42, 44] and V.V. Podinovsky, V.D. Nogin [72].
SUBJECTS: Organizer, Prognosist, Expert.
GOALS OF THE STAGE:
The main task of the stage is to give a conclusion concerning an aposteriori appraisal of the prognosis quality. In connection with this it is necessary:
1. To realize a verification of the prognosis [Organizer, Prognosist, Expert].
2. To get an information on an analogous, but outside, prognosis with the same forestalling period [Organizer, Prognosist, Expert].
3. Through a comparative analysis to expose what prognostic appraisal is more correct: obtained by ourselves or obtained within the frameworks of the outside prognosis [Prognosist, Expert].
DESCRIPTION OF THE STAGE:
After a forestalling period completion it is necessary to realize operation on a verification of the prognosis and on an appraisal of its quality. In a number of cases these operations may be quite complicated.
The main reasons which complicating the verification and the prognosis quality appraisal are falling into two classes:
1. Firstly, they are subjective reasons related to a desire of forecasters to conceal defects of the prognosis and its shortcomings.
2. Secondly, they are objective reasons related to difficulties in revealing the truth, to a lack of information on an analogous prognosis, and mathematical complications related to a comparative analysis by accuracy.
In order to draw a conclusion related to the prognosis quality aposteriori appraisal we should have three appraisals: our appraisal, a real value of the object’s appraised parameter and the expert appraisal with which we will compare the one which was accepted by ourselves.
While comparing two expert appraisals with a real one we are setting the problem to ascertain which of them is better (more accurate). Therefore, we may consider only those types of appraisals for which it is possible to introduce an order ratio in a natural way. Furthermore, only where it is possible to introduce a "between" ternary ratio.
In a general case the order discussed on a multitude may be a partial one, that is not all the elements could be compared between themselves by this ratio.
The conclusion concerning the best appraisal during a comparison may also depend on the fact what error index was used.
The problem of a comparative analysis by accuracy was both unsolved, and has not been raised up to now. At present time it remains actual. The debates concerning a better appraisal in case of a non-sufficient understanding of an argument subject are being continued in our days [149].
As an explanation let's consider the simplest case when an expert appraisal is a number and let's indicate the following error indexes:
where x, y, L, U ÎR, x - an expert appraisal, y – real value, L and U - the lower and upper limits of the scale shown to an expert, D = max {U-y, y-L}, 0<L<U.
The error indices shown are practically exhausting the list of positively defined error indices described in the literature on the expert appraisals (for numerical appraisals).
Let's show examples of the error indices when the appraisals are expressed in the form of vectors:
E n1(x,y) =S ni|xi - yi| and E n2 (x,y) = S ni (xi - yi)2,
where xi, yiÎR, i=1…n.
NOTE:
The introduction of error index notion in an axiomatic way, a sign equivalence of indices, and a proof of a number of the necessary and sufficient conditions and theorems allowed the author to solve partially the problem of a comparative analysis by the accuracy. The materials on this subject are described in detail in [79].
