Postupci analize podataka u izgradnji profila korisnika usluga

scientific master's theses

• Author: Filip Ujević
• Mentor: Nikola Bogunović (mentor)
• Committee member: Bojana Dalbelo-Bašić (predsjednik povjerenstva)
• Committee member: Nikola Bogunović (član povjerenstva)
• Committee member: Dragan Gamberger (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Electronics, Microelectronics, Computer and Intelligent Systems); Zagreb
• Defense date and country: 2004, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering
• Universal decimal classification (UDC): 621.3 - Electrical engineering

Abstract

Suvremene računalne i telekomunikacijske tehnologije omogućile su jednostavno prikupljanje i jeftinu pohranu ogromnih količina podataka. U poslovnom svijetu prikupljeni podaci su uglavnom transakcijskog oblika, a najveći dio nastaje bilježenjem interakcija s korisnicima usluga. Pravi potencijal tako izgrađenih baza podataka nije u sirovim transakcijskim podacima, već u implicitnim pravilnostima koje su u njima skrivene. U današnjem svijetu globalne konkurencije veliku ulogu u stjecanju i održavanju tržišnog udjela ima upravo pronalaženje, razumijevanje i iskorištavanje pravilnosti u obilježjima i ponašanju korisnika usluga. Postupci inteligentne analize podataka omogućuju strojno pronalaženje implicitnih pravilnosti i odnosa koji postoje skriveni u velikim bazama podataka. U ovom radu izložene su osnovne postavke i ideje na kojima se zasnivaju postupci inteligentne analize podataka, opisani su temeljni problemi u njihovoj primjeni, kao i načini rješavanja tih problema. Prikazane su najpopularnije tehnike svake od faza procesa inteligentne analize podataka, te su naznačena njihova svojstva i međusobne razlike. Posebice, rad istražuje adekvatnost, efikasnost i uspješnost primjene opisanih postupaka inteligentne analize podataka na stvarne probleme iz poslovnog svijeta. Analiziraju se konkretni podaci na tipičnom problemu profiliranja korisnika iz domene osiguravajućih društava. Provedena razmatranja rezultirala su specifičnom kombinacijom tehnika probabilističkog i deskriptivnog modeliranja, koja je prvenstveno namijenjena problemima s izraženim nedeterminističkim obilježjima. Na tim osnovama je izgrađen i programski produkt koji implementira postupke analize podataka u kontekstu promatranih problema.

Abstract (english)

Contemporary information and communication technology facilitates data acquisition and provides affordable storage for vast quantities of data. Business world resides on transactional data, and the largest part of transactional data comes from recording interactions with customers. The true potential of transactional databases is not in raw transactional data, but in implicit regularities that underlie the data. In the modern world of global competition, discovering, understanding and exploiting patterns in customer characteristics and behavior has an important role in obtaining and maintaining substantial market share. Data mining techniques enable automatic extraction of implicit regularities and hidden relationships that exist in large databases. This thesis explicates fundamental principles and ideas that form the basis of data mining techniques, explains essential problems in their practical application, and offers standard solutions to these problems. The most popular techniques for each phase of data mining process are also described, along with their properties and important differences. Particularly, the thesis explores adequacy, efficiency and usefulness of practical application of the described data mining techniques on real-world business problems. As a typical example, a problem of constructing customer profiles is taken from the insurance business. Conducted analysis and experiments resulted in a specific combination of probabilistic and descriptive modeling techniques, primarily suited for problems that exhibit a high degree of undeterministic behavior. Derived from these results, a computer application was developed, implementing data mining techniques in the context of target problem type.

• Keywords: inteligentna analiza podataka; otkrivanje znanja; strojno učenje; profili korisnika usluga; stvarni poslovni problemi; tehnike modeliranja; neizvjesnost
• Keywords (english): data mining; knowledge discovery; machine learning; customer profiles; real-world business problems; modeling techniques; uncertainty
• Language: croatian
• URN:NBN: urn:nbn:hr:168:759449
• Study programme: Title: Postgraduate master programme in electrical engineering; Study programme type: university; Study level: postgraduate; Academic / professional title: Magistar znanosti elektrotehnike (Magistar znanosti elektrotehnike)
• Catalog URL: https://lib.fer.hr:8443/cgi-bin/koha/catalogue/detail.pl?biblionumber=29334
• Type of resource: Text
• Extent: iii, 137 str. ; 30 cm
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-03-02 13:56:24