Metody predykcji w analitycznym Consumer Relation Management na potrzeby marketingu relacji

Close contacts with the customer, analysis of customer service costs and revenues generated by them, databases and information technology have become the prerequisites for an effective marketing orientation. Analysis cannot do without models referring to the conditions of companies’ activities. Models of the instruments of market influence allow to predict the results of the planned activities in the changing market environment. For these reasons, much attention is paid to the design and formal verification of empirical models optimising marketing communication addressed to the client in order to develop the client’s value for the company. The paper reviews the formal models used to optimise the selection of target groups for marketing communications in analytical Customer Relation Management. The effectiveness of parametric, non-parametric and semi-parametric predictive tools is discussed. The paper also considers a quantitative approach to the optimisation of target groups among Polish households that was to participate in the scientific project entitled “Determiners of decisions concerning education”. The hybrid approach was also described. In this case, this kind of approach consists in combining data mining tools (classification tree) with other analytical tools (logistic regression).