KNOWLEDGE DISCOVERY PROCESS FOR BUILDING CUSTOMER PROFILES PROCES OTKRIVANJA ZNANJA ZA IZGRADNJU PROFILA KUPACA

The knowledge about customer preferences and behavior is fundamental for personalization of products and service. Personalization products and services are possible only if we have enough knowledge of who customers are, how they are similar among, how they behave. Knowledge discovery is process of transforming data into knowledge by adequate algorithms and software tools. In the paper is developed an approach that uses data in the form of transactional databases to construct accurate individual profiles. In developed data model are integrated transactional data and rules describing customer’s behavior. The rules are extracted from transactional data and cover individual customer behavior as well as the common behavior of all customers in the market segment. There are two rules types: first, for describing individual customer behavior and second, for the market behavior. Knowledge discovery plays a crucial role as an enabler to the organizations to integrate effective analytical data mining methods for prediction, classification, cluster, anomaly detection with data management and information visualization. Knowledge discovery is oriented to learning. In the process of learning we are implementing the functions of R language and this tool has shown satisfactory application and development power. Sažetak Znanje o ponašanju kupaca i njihovim preferencijama je ključno za personalizaciju proizvoda i usluga. Personalizacija proizvoda i usluga je moguća samo ako imamo dovoljno znanja o tome tko su kupci, koliko su međusobno slični, kako se ponašaju. Otkrivanje znanja je proces transformacije podataka u znanje primjenom odgovarajućih algoritamai softverskih alata. U radu je razvije pristup koji koristi podatke u obliku transakcijeke baze podataka kako bi se izgradio individualni profil. Pravila se ekstarhiraju iz transakcijskih baza podatakai otkrivaju pojedinačno ponašanje kupaca ali i njihovo zajedničko ponašanje unutar nekog klastera. Zato postoje pravila koje otkrivaju pojedinačno ponašanje kupca ali i cijelog tržišta (klastera). Otkrivanje znanja je tako orijentirano učenju. U procesu učenja primjenili smo fnkcije jezika R i taj softverski alat je pokazao zadovaljavajuću aplikacijsku i razvojnu moć.