A dynamic linear model to forecast hotel registrations in Puerto Rico using Google Trends data

Abstract

Recently, studies have used search query volume (SQV) data to forecast a given process of interest. However, Google Trends SQV data comes from a periodic sample of queries. As a result, Google Trends data is different every week. We propose a Dynamic Linear Model that treats SQV data as a representation of an unobservable process. We apply our model to forecast the number of hotel nonresident registrations in Puerto Rico using SQV data downloaded in 11 different occasions. The model provides better inference on the association between the number of hotel nonresident registrations and SQV than using Google Trends data retrieved only on one occasion. Furthermore, our model results in more realistic prediction intervals of forecasts. However, compared to simpler models we only find evidence of better performance for our model when making forecasts on a horizon of over 6 months. IntroductionIn recent years, there has been an interest in exploiting search query data available through sources such as Google Trends (www.google.com/trends) to model temporal processes. Choi andVarian (2012, 2009) used search query data to model tourism demand, auto sales, home sales, and initial unemployment claims. Ginsberg et al. (2009) relied on Google search queries to model influenza activity in the U.S. Studies have also suggested search query based tools to model consumer behavior (Goel et al. 2010), dengue (Gluskin et al. 2014) and more. Similarly, Yang et al. (2014 used web traffic data to model hotel demand. It is not exactly known how the search query volume algorithm by Google generates its results. Moreover, the time series of search query volume generated by the algorithm changes every week.Puerto Rico has been going through an economic recession since 2006. Leaders on the island have been attempting to find ways to boost the economy. Although hotel registrations from July to November

Modeling excess deaths after a natural disaster with application to Hurricane Maria

“…For Model 2, we consider US Bureau of Transportation Statistics data on monthly net movement of air passengers to adjust population estimates (Table ). Usually, net movement is a biased proxy of resident migration mainly because it includes the seasonal movements of visitors, and tourism has been one of the few growing sectors in the Puerto Rican economy . However, as Table shows, a dramatic drop in visitors from the US occurred the months following Maria's landfall (the US dominates the local tourism market share).…”

Incorporating Open Data Into Introductory Courses in Statistics

“…There are some challenges to be aware of when incorporating open data into the classroom. Sometimes it is not clear where the data comes from, which puts the reliability of the data in question ([Rivera(2016)]); or there is no variable dictionary available with the data set. Depending on the data set and context of a case study, parameters can be obtained, not statistics (e.g.…”

Spatial Configuration and Online Attention: A Space Syntax Perspective

“…Scholars have begun to utilize search engines [22,23], user-generated content (UGC) [24,25], user-generated photography [25], heat maps [16] and other Internet big data to study tourist preferences and spatial behavior [26,27], drawing from the rich data on geographical locations that are increasingly available. The studies based on big data mainly present opinions about landscape preference and posit that the majority of landscape users judge the value of the landscape based on their respective preferences.…”