Tracking customer portrait by unsupervised classification techniques
LTStraipsnyje analizuojami marketingo rodikliai, nusakantys klientų informacijos ir įmonės finansinių rezultatų sąsajas. Pasiūlytas kliento dinaminio portreto konceptas ir sudarytas analitinis modelis, padedantis nustatyti esminius klientų klasterizavimo kintamųjų rinkinius. Klasterizavimo modelis sudarytas remiantis neuroninių tinklų, kintamųjų jautrumo analize bei taikant savitvarkių žemėlapių intelektinius modelius. Modelis eksperimentiškai patikrintas naudojant turizmo agentūros pardavimų duomenis. Apibendrinus eksperimentinio tyrimo rezultatus, įvertintas kliento dinaminio portreto skaitinio įvertinimo efektyvumas prognozuojant finansinius rezultatus ir klientų migravimą tarp klasterių.
ENThe problem of the research is targeted to exploring the customer-related information by analysing marketing indicators in order to substantiate the enterprise financial results. The concept of dynamic customer portrait is introduced for creating analytical model The suggested model explores the most influential variable sets for identifying customer clusters and basis for their membership. The computational methods of neural network, sensitivity analysis and self-organized maps for unsupervised classification were applied and verified by the experimental research. The experimental research was performed by applying the suggested model for customer database of the travel agency. The results of the analysis were summarized and the research insights presented by analysing the effectiveness of the method in forecasting financial outcomes related to customer mapping and migrating between dusters over the dynamic development of the customer portrait indicators.
