Forecasting Hotel Occupancy Rates in Augusta: Can Google Trends Improve Forecasts?

Abstract

This project will develop models in an attempt to develop better forecasts of hotel occupancy for the market in Augusta, Georgiaby utilizing historical occupancy data and Google trends data. Using the historical data from the years 2012 through 2015, a series of five univariate modelswill be made with differing forecasting equations to forecast the year 2016. The forecast for theyear 2016 will be compared to actual occupancy data from 2016 to measure for errors. The models will then be re-estimated with additional keywords that will be chosen on the basis that they will be commonly used to search for and book hotels. Some terms will be specific to Augusta and others will be general for booking hotels. With those terms, an index will be created to weigh the terms according to their relevance throughout the year, according to Google trends. With the addition of the keywords, the newforecasts will be compared to actual occupancy data from 2016. Errors of the univariate models and the models utilizing Google trends data will be compared to determine the accuracy of the two forecasting techniques.