|
Time Series Intervention Analysis on Count Data of Foreign Tourist Arrivals with Log-Linear Poisson AR and Poisson Autoregressive (PAR(p)) a) Department of Statistics, Institut Teknologi Sepuluh Nopember, Surabaya 60111, Indonesia Abstract Number of foreign tourist arrivals data are count time series containing discrete values in the form of event data, which may follow the Poisson distribution. This type of count data is modeled using time series models like Log-Linear Poisson Autoregressive and Poisson Autoregressive (PAR(p)). The Log-Linear Poisson Autoregressive was developed using the GLM (Generalized Linear Model) method, while PAR(p) was developed using a state-space technique. Number of inputs known as interventions, such as natural disasters, pandemics and its supporting policy, have impacts on foreign tourist arrivals. The first goal of this study is to compare the three methods-Gaussian-AR, Log-Linear Poisson AR, and PAR(p)-to be used for forecasting and examining the effects of the interventions on the number of foreign tourists entering Indonesia through Ngurah Rai Bali Airport. The next goal is to calculate the potential loss that may be incurred based on the best method. COVID-19 pandemic is used as the main intervention in this paper. By contrasting the AIC, RMSE, MAD and MAPE values for each candidate model, Log-Linear Poisson AR was chosen as the model that best fits the data. According to the model, COVID-19 has been discovered to considerably reduce the number of tourists arriving at the Ngurah Rai Airport by 96.4% on average. The estimated values of potential losses due to the presence of COVID-19 pandemic intervention are determined to be 6,890,932,320.99 US dollars in the 2020-2021 biennium based on the monthly tourist expenditures per visit in the year of 2019. Keywords: Intervention analysis, Log-linear Poisson AR, PAR(p), Foreign tourist arrivals, COVID-19 Topic: MATHEMATICS AND STATISTICS |
| ICMNS 2023 Conference | Conference Management System |