Definition and evaluation of forecast quality models for the integrated circuits with the use of theory of pattern recognition

The paper is devoted to the technique of the forecast models for the integrated circuits of 564 series based on the methods of the theory of pattern recognition. Methods of regression models and discriminant functions were used for the model-building. Analysis of the obtained models was carried out. The optimal threshold values of the discriminant and regression functions were determined. The values of the risk of incorrect decision, producer’s risk (α-risk) and consumer’s risk (β-risk) were estimated.