S.V. Vasilyev
SOLERC, P.O. Box 59, Kharkiv, 310052, Ukraine
Much of the astronomical data analysis proceeds from the aforegoing observational material or theoretical predictions. However, in some instances, especially for the first-observed assemblage of curves, it may be efficient to process and classify the data without a priori assumption on the objects under study. A technique allowing to describe data automatically by relatively small number of independent parameters based on the principal component analysis of data sequences will be presented and some examples will be given. Among the advantages of the technique belong stability of the obtained parameters to further adding new data and minimizing the rms errors of data representation. Our software based on this approach affords a serious ground for creating uniformal data systems and allows to fill gaps in the truncated data records. Strong and weak points of the proposed technique as well as its basic requirements to the initial data structure and hardware will also be discussed. This work was supported by the ISSEP through grant N PSU052018.