Time series analysis and prediction using case based reasoning technology. Analiza i predviđanja toka vremenskih serija pomoću "case-based reasoning" -tehnologije

Abstract

This thesis describes one promising approach where a problem of time series analysis and prediction was solved by using Case Based Reasoning (CBR) technology. Foundations and main concepts of this technology are described in detail. Furthermore, a detailed study of different approaches in time series analysis is given. System CuBaGe (Curve Base Generator) - A robust and general architecture for curve representation and indexing time series databases, based on Case based reasoning technology, was developed. Also, a corresponding similarity measure was modelled for a given kind of curve representation. The presented architecture may be employed equally well not only in conventional time series (where all values are known), but also in some non-standard time series (sparse, vague, non-equidistant). Dealing with the non-standard time series is the highest advantage of the presented architecture.

U ovoj doktorskoj disertaciji prikazan je interesantan i perspektivan pristup rešavanja problema analize i predviđanja vremenskih serija korišćenjem Case Based Reasoning (CBR) tehnologije. Detaljno su opisane osnove i glavni koncepti ove tehnologije. Takođe, data je komparativna analiza različitih pristupa u analizi vremenskih serija sa posebnim osvrtom na predviđanje. Kao najveći doprinos ove disertacije, prikazan je sistem CuBaGe (Curve Base Generator) u kome je realizovan originalni način reprezentacije vremenskih serija zajedno sa, takođe originalnom, odgovarajućom merom sličnosti. Robusnost i generalnost sistema ilustrovana je realnom primenom u domenu finansijskog predviđanja, gde je pokazano da sistem jednako dobro funkcioniše sa standardnim, ali i sa nekim nestandardnim vremenskim serijama (neodređenim, retkim i neekvidistantnim).