Nowcasting RDU with trends Based on Durham Paper By Ramy Khorshed
About Google Trends Google Search query volume Y-axis search index X-axis time In 2008, Google launched Google Insights for Search Revamped front-end in 2012
Google Trends: Example Lax Scandal Jane Goodall Primate Center Steve Jobs Speech
Google Trends: Example
Proof of Concept: Etteredge (2005): US unemployment rate Cooper (2005): Cancer Polgreen(2008) and Ginsberg (2009): Contagious diseases Choi and Varian (2009): Unemployment Automobile demand Vacation Destinations Goel (2010): Box-office revenue First Month sales of video games Rank of songs on the Billboard Hot 100
Durham Paper Topic: Can applying simple regression models enhanced by Google search volume data can improve the predictability of current and near-future economic conditions pertaining to Durham? Specifically, I will adjust predictions of Raleigh-Durham International (RDU) passenger volume based on the number of queries related to RDU.
Methodology Model 0: log(y t ) = α 1 log(y t-1 )+ α 2 log(y t-12 )+e t Model 1: log(y t ) = α 1 log(y t-1 )+ α 2 log(y t-12 )+ α 3 x t +e t Data:
Methodology Model 0: log(y t ) = α 1 log(y t-1 )+ α 2 log(y t-12 )+e t Model 1: log(y t ) = α 1 log(y t-1 )+ α 2 log(y t-12 )+ α 3 x t +e t Trend Data:
Results: MAE = (1/T) T t=1 |Pe t | model 0 = 4.35% model 1 = 3.31% Improvement of 31.41%
Conclusions: This result could help airport management better predict passenger volume allowing them to make better decisions and improve customer experience. Durham hotels could look to more accurately anticipate demand for lodging and accordingly change price by incorporating search volumes into predictions based on past occupancy. Durham real estate developers could incorporate monthly and daily query volumes for Durham to help determine real- estate value. Raleigh-Durham searches from the search could be used to help guide marketing decisions.