Instruments and Systems: Monitoring, Control, and Diagnostics Annotation << Back Transformations on a generalized axis and analysis of their quality in the problems of diagnosis and pattern recognition SUBBOTIN S.A. The actual problem of software development for data dimensionality reduction is decided in the paper. The set of the rapid transformation of the original multi-dimensional space to a one-dimensional axis has is firstly proposed. It is based on the principles of hashing, and shall take into account the example locations in the feature space relative to the centers of mass of classes, and also allows to evaluate and take into account the weights of the features and implicitly solves the feature selection problem. A set of indicators that characterize the properties of the transformations from the multidimensional space to a generalized axis is firstly proposed. It allows to compare and choose the best transformation and the best results of the transformations in the solving of diagnosis and recognition problems on the basis of precedents. An experimental investigation of the proposed transformations and indicators held in solving problems of diagnosis and pattern recognition. Its results allow to recommend the proposed transformations for use at the practice. Keywords: sample, dimensionality reduction, feature extraction, transformation, generalized axis, feature selection, diagnostics, pattern recognition. Contacts: E-mail: subbotin@zntu.edu.ua Pp. 08-16.