A Markov Chain Model to Analyze the Entry-and-Stay States of Frequent Visitors to Taiwan

A model to predict the immigration behaviors of frequent visitors would help to improve clearance services and resource allocation at a country's border. This research uses Markov process to analyze the entry-and-stay states of frequent visitors based on their immigration records. Prior studies have lacked quantitative information about the entry and stay states of travelers at the border. In this study, the following attributes were drawn from the immigration records: (1) entry and exit date, (2) entry and exit frequency, and (3) duration of stay. We calculated a transition probability matrix containing all transition probabilities between each entry-and-stay states of visitors. When entry event of a visitor occurs in a certain state, we can estimate the possible state in the next period and the equilibrium probability by using the transition matrix. We determines the transition state of visitors entering Taiwan, and to consider the overall transition probabilities to predict the immigration behaviors. The model results in the steady-state probability. The state S5 (Entering 2 to 8 times and staying 3 to 6 days) has the highest probability of 27.99%. The definition of frequent visitor can be revised by the implication of state S5 to improve future decisions and immigration services based on these results.