Table 7.
ARIMAa/ARIMAXb model performance (RMSEc) for New York–Newark–Jersey City, NY-NJ-PA, Philadelphia-Camden-Wilmington, PA-NJ-DE-MD, Phoenix-Mesa-Chandler, AZ, and Washington-Arlington-Alexandria, DC-VA-MD-WV. Models that performed better than the baseline ARIMA are shown in italics.
| New York–Newark–Jersey City, NY-NJ-PA, RMSE | Philadelphia-Camden-Wilmington, PA-NJ-DE-MD, RMSE | Phoenix-Mesa-Chandler, AZ, RMSE | Washington-Arlington-Alexandria, DC-VA-MD-WV, RMSE | |
| (Baseline) percentage of individuals who have been administered at least 1 vaccine dose (7 day rolling average) | 4.7686 | 4.4737 | 2.7743 | 2.5829 |
| Number of users per 100,000 population | 1.8473 | 1.2524 | 1.9983 | 1.1554 |
| Number of tweets per 100,000 population | 1.8329 | 1.2737 | 1.9769 | 1.1259 |
| Average favorites | 4.7564 | 5.6078 | 1.9406 | 0.7570 |
| Average retweets | 1.8838 | 0.9200d | 2.7728 | 0.7570 |
| % Positive sentiment, | 1.8682 | 4.8339 | 1.9452 | 1.1168 |
| % Negative sentiment | 1.8757 | 1.2486 | 1.9118 | 1.1206 |
| % Neutral sentiment | 4.7722 | 1.2392 | 1.8932 | 2.5825 |
| % Trust | 1.8659 | 1.2503 | 1.9372 | 1.1327 |
| % Surprise | 4.7668 | 1.2279 | 1.9374 | 1.1210 |
| % Sadness | 4.4896 | 1.1615 | 1.9355 | 2.4392 |
| % Joy | 1.8397 | 1.1956 | 1.9424 | 1.1450 |
| % Fear | 4.7720 | 4.5114 | 1.9371 | 1.0632 |
| % Disgust | 1.8207d | 1.2506 | 1.9520 | 1.1380 |
| % Anger | 1.9003 | 4.6179 | 1.8858d | 0.6834 |
| % Anticipation | 1.9060 | 1.2348 | 1.9454 | 1.1088 |
| Best predictors: joy (%), negative sentiment (%), surprise (%), trust (%) | 2.7323 | 33.5446 | 5.1538 | 0.6816d |
aARIMA: autoregressive integrated moving average.
bARIMAX: autoregressive integrated moving average exogenous variable model.
cRMSE: root mean square error.
dBest-performing model for each metropolitan area.