o:2796 Geometrijski minifikacioni vremenski nizovi generisani modifikovanim negativnim binomnim operatorom Geometric minification time series models generated by the modified negative binomial operator : doctoral dissertation sr This thesis introduces new minification INAR(1) models. One-dimensional and two-dimensional models are introduced. The models are based upon the modified negative binomial operator. The most important statistical properties of the models are determined. Model parameter estimation is performed using various methods. All the methods are tested on a simulated dataset. An application to a real dataset is presented. Special attention is devoted to the analytical determination of parameter estimates for the one-dimensional minification INAR(1) model, via the application of the maximum likelihood method. An EM algorithm is constructed. The quality of estimates and the speed of the algorithm are tested on a simulated dataset. Biografija: listovi 110-111Bibliografija: list [112] Datum odbrane: 23.09.2024. Time series analysis 91552100 155557385 91552101 8731 2024-12-30T09:24:37.053Z 45 no 46 Milena, 1993- Stojanović 119246857 2024 63 mentor Miroslav, 1972- Ristić 13816935 2024 63 član komisije Maja S., 1980- Obradović 31877223 2024 63 član komisije Miodrag, 1974- Đorđević 11556455 2024 63 član komisije Predrag, 1982- Popović 14041959 2024 [6], 111 listova 2214573 http://phaidrani.ni.ac.rs/o:2796 no yes 12 2024-12-30T09:24:37.320Z 70 minifikacioni modeli, celobrojni autoregresivni modeli, modifikovani negativni binomni operator, geometrijska marginalna raspodela, EM algoritam minification models, integer-valued autoregressive models, modified negative binomial operator, geometric marginal distribution, EM algorithm 519.246.8(043.3) P160 1738 18A07 18A0701 2024