The COVID-19 pandemic is greatly affecting the hospitality industry worldwide. Lodging demand is severely reduced as people's fear of coronavirus spreading risk in hotels. This research makes a timely contribution to the hospitality literature by proposing the mixed data sampling models (MIDAS) to monitor and forecast latest hotel occupancy rates with high-frequency big data sources, such as daily visitor arrivals and search query data. Quantitative evidence from Macau from January to July 2020 confirms that MIDAS models can measure the dynamic impacts of the COVID-19 pandemic on hotel occupancy and have a better prediction accuracy and explanation ability than competitive models. Industry practitioners can adopt this big data analytical framework to make daily or monthly updates of lodging demand, conduct scenario analysis, plan and trace the recovery schedule during and post COVID-19 phases. Finally, managerial implications and future work are highlighted.