Table 2.
Accuracy in predicting the number of new HIV diagnoses in China
| Model | Nowcasting | One-month ahead forecasting | Two-month ahead forecasting | |||
| RMSE | NRMSE | RMSE | NRSME | RMSE | NRMSE | |
| nbGLM-AR | 473.35 | 11.71% | 484.21 | 11.98% | 528.38 | 13.08% |
| nbGLM-Baidu | 957.58 | 23.7% | 1166.38 | 28.86% | 1176.06 | 29.1% |
| nbGLM-AR-Baidu | 420.68 | 10.41% | 482.79 | 11.95% | 539.37 | 13.35% |
| BnbGLM-AR | 455.65 | 11.27% | 456.95 | 11.31% | 497.11 | 12.3% |
| BnbGLM-Baidu | 976.99 | 24.18% | 1176.16 | 29.11% | 1145.23 | 28.34% |
| BnbGLM-AR-Baidu | 423.17 | 10.47% | 451.75 | 11.18% | 508.31 | 12.58% |
BnbGLM-AR, Bayesian negative binomial generalised linear model (BnbGLM) with autoregressive terms; BnbGLM-AR-Baidu, BnbGLM with autoregressive terms and the composite Baidu Search Index; BnbGLM-Baidu, BnbGLM with a variable representing the composite Baidu Search Index; nbGLM-AR, negative binomial generalised linear model (nbGLM) with autoregressive terms; nbGLM-AR-Baidu, nbGLM with autoregressive terms and the composite Baidu Search Index; nbGLM-Baidu, nbGLM with a variable representing the composite Baidu Search Index; NRMSE, normalised root mean square error; RMSE, root mean square error.