Table 3. Calculation of key accuracy indexes spanning 1994–2009 evaluating the predictive quality of the D-R and GM(1,1) models.
I-U | I-G | D-U | D-G | |||||
D-R | GM(1,1) | D-R | GM(1,1) | D-R | GM(1,1) | D-R | GM(1,1) | |
SSE | 2.76 | 3.58 | 21.43 | 49.54 | 0.01 | 0.01 | 0.04 | 0.38 |
RSE | 0.43 | 0.49 | 1.20 | 1.82 | 0.02 | 0.03 | 0.05 | 0.16 |
MAD | 0.28 | 0.40 | 0.81 | 1.29 | 0.02 | 0.02 | 0.04 | 0.08 |
ARE | −1.26% | 4.65% | −1.50% | −13.22% | −1.70% | 6.39% | −0.18% | −12.97% |
MAPE | 3.99% | 6.59% | 7.93% | 13.22% | 6.46% | 7.65% | 7.86% | 14.17% |
Note: I-U, I-G, D-U and D-G represent the incidence (I) and deaths (D) of tuberculosis in the US (U) and Germany (G), among HIV-negative TB patients in the. D-R and GM(1,1) define the algorithms from which these values were derived.