Interactive Algorithms for Expert Evaluation of Projects of Reconstruction of Urban Districts

摘要

As an example, we show the scheme of a committee of ten recognizing systems which allows us to recognize (after learning) the estimates of new projects according to a ten-point evaluation scale. Each system can be constructed according to an algorithm where the threshold of argument efficiency is established subjectively by the author of simulation or objectively according to the combinatorial algorithm of GMDH (group method of data handling), where the optimal model is determined by the external criterion computed on a separate test data sample. In a similar way, subjectively or objectively, one can obtain the expert's estimations. One common group of experts can give only subjective judgments. For the objective self-organization of projects, experts must be divided into learning and test groups. The functions of each of these groups are indicated. With a ten-point scale of evaluation of the project quality, one can obtain ten discriminant models for estimating the projects that were not considered by experts.