Identification model based on latent variables
С. А. Баркалов, Olga Bekirova, N.Yu. Kalinina, Sergey I. Moiseev · Journal of Physics Conference Series · 2020
Abstract The paper proposes a model for assessing the degree of conformity of an object to a standard type. Evaluation is based on criteria based on identification signs. The model is based on the Rasch’s method of estimating latent variables. This is due to the fact that the degree of belonging of the evaluated objects to the sample is a latent variable. An original method is proposed, which assumes that indicators that allow identification are distributed according to the normal law. This allows us to estimate the average deviations of indicators from a given value and, taking into account the variance, to assess the degree of belonging of the object to the original. The obtained estimates are compared with the estimates obtained by traditional methods. Using computational experiments, it is shown that the model provides adequate estimates and can be used for identification. The method of implementation of computational procedures in MS Excel is described. The advantages of the described identification model are given in comparison with traditional models that solve this problem.