Modeling of tourist profiles with decision trees in a world heritage city: the case of Cuenca (Ecuador)

Abstract

This research project proposes the formulation of profiles composed of foreign and national tourists in a world heritage city as a contribution to the strategies of the Association of Accommodation Establishments in the city of Cuenca. A total of 1,293 surveys, gathered during holidays between 2015 and 2016, underwent a statistical analysis of decision trees and statistical regression using seven variables. In addition, by means of CART and Random Forest technical comparisons, the models and predictive variables were prioritized for marketing strategies and planning. Two tourist profiles were described, the backpacker type and the cultural type, with an error of 1%, where the variables of visitor age, daily average accommodation expense, and type of employment at the origin predicted the accommodation selection upon arrival at the tourist destination in a better way.