Author(s)

A. RABASA, A. PÉREZ-MARTÍN & D. GINER

Abstract

The Valencian Community (South-East Spain) is one of the most important tourist destinations in Europe. The Valencian Government has been carrying out surveys about the types of travel, the type of transport, the type of accommodation, the duration of the trip and the number of travellers, as well as other issues. The aim is to discover the different spending typologies incurred by foreign visitors.

In their task of drawing up more attractive tourist strategies, the following questions may become particularly relevant to the Valencian Public Services: what type of traveller spends more on transportation in their own country, or pays for it in the Valencian Community; visitors’ nationalities and their higher or lower propensity to spend money on leisure; or the number of overnight stays in low-end destinations.

But the surveys gathering all this information consist of multiple and nested responses, distributed in thematic blocks that overlap, and whose translation to flat file systems (susceptible to being analysed with acceptable counting times) is a complex problem.

This paper presents a treatment process of the surveys, especially oriented towards having a suitable dataset to generate models of optimal segmentation of the different types of expenditure. Likewise, some results of such segmentation are shown, which are proving to be of great value to public managers in their challenge to offer suitable tourist alternatives to each type of traveller.

The paper includes an example of how open data sources can be incorporated into the original dataset in order to obtain better segmentation. A variation to the classical segmentation methods (algorithms of the K means family) is also provided, which leads to the establishment of the optimal number of groups for each computational experiment.

Keywords

Big data, clustering, optimization, surveys analysis, tourism