TY - THES ID - cogprints5625 UR - http://cogprints.org/5625/ A1 - JÄNTSCHI, Dr. Lorentz Y1 - 1995/09// N2 - The Thesis is the dissertation for Licensed in Informatics Degree. The thesis threat the subject of time series from point of view of decomposing the knowledge from a time series into trend, ciclicity, and periodicity. All this components are modeled separately and a software application embeds all thee in a compact ensamble. The application is useful in prediction. A set of applications were taken into consideration and analyzed using this methodology. Applications are from meteorology, demographics, chemical and physical properties measurements. The software application was writen in Turbo Pascal and contain a statistical kernel, a graphical kernel and a user interface which allows specifying of the choused model. Trends implemented are linear, polynomial, hyperbolic, logarithmic and exponential. The periodicity module contains a autocorrelation procedure and a Fourier analysis procedure. PB - "Babes-Bolyai" University KW - Time series; Prediction; Cyclicity; Periodicity; Trend; Correlation; Ranks correlation; Self correlation; Harmonic series; Fast Fourier transformation TI - Time Series. Prediction AV - public ER -