CONSUMERS CLASSIFICATION ACCORDING INCOME BY USING NEURAL NETWORKS AND DECISION TREES

Rad se bavi kreiranjem i analizom modela neuronskih mreža i stabala odlučivanja u svrhu klasifikacije potrošača prema prihodu. Ulazne varijable na kojima je model kreiran su: starost, vrsta zaposlenja, stupanj obrazovanja, broj godina obrazovanja, bračni status, zanimanje, status veze, rasa, spol, prihod kapitala, gubitak kapitala, broj radnih sati u tjednu te zemlji porijekla. Izlazna varijabla kreiranog modela je godišnji prihod. U radu je napravljena usporedba uspješnosti modela neuronske mreže i modela stabla odlučivanja. Kreirane modele, uz dodatno treniranje na novijim podacima moguće je koristiti za planiranje marketinških strategija pri izradi promocijskih aktivnosti različitim skupinama potrošača sukladno njihovom prihodu.The Graduate paper deals with the creation and analysis of neural networks and decision trees in order to classify customers. The input variables on which model was created are: age, workclass, education, years of education, marital status, occupation, relationship, race, gender, capital gain, capital loss, hours per week, native country. The output variable on created model is income. Comparison of the performance was made in the neural network model and decision tree model. Created models, with additional training on newer data, can be used to plan marketing strategies when developing promotional activities for different consumer groups according to their income