bokangntovana/SA-Tourist-Accommodation-Forecasting

This project develops a robust and operationalized time series model to forecast the Total Industry Occupancy Rate for South African accommodation establishments, using historical data from Stats SA (2007–2025). # SA-Tourist-Accommodation-Forecasting # Project Overview This project develops a robust and operationalised time series model to forecast the Total Industry Occupancy Rate for South African accommodation establishments, using historical data from Stats SA (2007–2025). The core challenge was to accurately model the series' strong seasonality while maintaining accuracy during and after the extreme structural shock caused by the COVID-19 pandemic. A hybrid approach comparing classical time series modelling (SARIMAX) with non-linear Machine Learning (XGBoost) was used.