Space and Time Analysis of Tourist Movements using Semi-Markov Processes

Abstract

Tourist movement is a complex process, but it provides very useful information for park managers and tourist operators. This paper aims to establish a sound methodology for modelling the spatial and temporal movement of tourists, with the objectives of understanding, predicting, controlling and optimising the decisions made by them as they go about choosing the attractions they want to visit. Tourist movements, in this paper, are modelled as discrete processes between specific tourist locations, which could be located some distance apart. A Semi-Markov process has a Markov chain and a renewal process embedded within its structure. Therefore, the Semi-Markov chain can be used to understand the interaction of tourists with attractions as a sequence of movements over time, rather than their interaction with individual attractions. The following two assumptions, which underlie the Semi--Markov process, make it an especially ideal tool for modeling movements of tourists: The probability that a tourist will visit a particular attraction depends only on the most recent attraction that was visited by that tourist. The distribution of the time spent at each attraction is dependent on both that attraction and the next attraction that is visited.One of the outcomes of this approach is a measure which assesses the attractiveness of particular tourist attractions based on spatial and temporal interactions between the attractions. Two assumptions, based on the assumptions of the Semi--Markov process, are derived for assessing attractiveness of tourist attractions. The more tourists visit an attraction, the more attractive it is. A transition probability matrix is developed for estimating the probability that a tourist will visit a particular attraction based on the first assumption. The longer tourists stay at an attraction, the more attractive it is. A mean time transition matrix is calculated, based on the second assumption, to estimate the time spent at each attraction. The attractiveness of each attraction can then be calculated based on these two matrixes. A case study conducted at Phillip Island Nature Park, Victoria, Australia is used to validate the model. The studies’ results prove that model is efficient. They are also useful, in that knowing which attractions are the most popular, how long tourist will spend at any one site, and what the likely routes are that they will follow and how attractions associate with each other, can inform marketing decisions of park managers and tourist operators.