Time Series Regression

作者：Michael Campbell, Richard Jacques · 年份：2023 · DOI：10.1002/9781119401407.ch8 · 研究领域：COVID-19 epidemiological studies、Forecasting Techniques and Applications、Air Quality Monitoring and Forecasting

Time series regression is mainly used when the outcome is continuous, but measured together with the predictor variables serially over time. The potential for confounding in time series regression is very high – many variables either simply increase or decrease over time, and so will be correlated over time. Potential confounding factors are seasonality and trend in the model. The walk-in centre was closed at night and so was not expected to affect the night-time emergency department attendances. The pandemic was declared in March 2020 and so monthly health data from January 2018 to February 2020 were used as a baseline against which to compare utilisation rates for April–June 2021. Auto-correlation was controlled by performing a Durbin–Watson test to test the presence of first-order auto-correlation and because auto-correlation was detected, using the Prais–Winsten generalised least squares estimator to estimate the regression coefficients.