U.S. Tourist Arrivals & Spending Forecasting
Function-of-time and SARIMA modeling of U.S. monthly tourist arrivals and spending (2000–2024), validated by out-of-sample forecasting a full holdout year.
TECH STACK
Deep Dive
Time series analysis of U.S. monthly international tourist arrivals and tourist spending ("travel exports") from 2000–2019 (pre-pandemic), sourced from the International Trade Administration's I-94 Arrivals and Travel Receipts/Payments programs.
Compared function-of-time regressions (linear, quadratic, seasonal dummies, cosine terms) against SARIMA models selected via ACF/PACF-guided differencing and AIC/BIC, with residuals validated for stationarity, independence, and normality at each step.
The selected SARIMA(0,1,1)(0,1,1)[12] model for arrivals validated strongly out-of-sample, forecasting the 2019 holdout year with R² = 0.97, and was used to project a counterfactual 2020–2021 absent COVID-19 disruption. A parallel SARIMA(0,1,0)(0,0,1)[12] model was fit for tourist spending.