COVID-era forecasting: Google trends and window and model averaging
研究了在历史数据不适用时，利用谷歌趋势数据结合窗口和模型平均方法改进旅游需求预测，以爱丁堡为例验证了效果。
• Google Trends data can provide extra information when historic data are unsuitable. • Informative Google Trends data may show short-term fluctuations in unstable periods. • Model averaging using recent performance adapts to short-term changes in relevance. • Window averaging helps to mitigates against uncertainty in the relevance period. • Edinburgh tourism demand forecasts are improved by such window and model averaging.