Table 4:
Comparison of Forecast Performance Based on Empirical Data (N = 142)
| Null Model | ZIMLP | RS-ZIMLP | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Mean | SD | 95% CI | Mean | SD | 95% CI | Mean | SD | 95% CI | |
| ACC* | 0.66 | - | - | 0.62 | 0.01 | [0.60, 0.64] | 0.82 | 0.01 | [0.80, 0.85] |
| recall* | 0.00 | - | - | 0.73 | 0.02 | [0.69, 0.76] | 0.77 | 0.03 | [0.72, 0.83] |
| precision* | - | - | - | 0.46 | 0.01 | [0.45, 0.48] | 0.71 | 0.02 | [0.68, 0.75] |
| AUC | - | - | - | 0.69 | 0.01 | [0.66, 0.71] | 0.82 | 0.01 | [0.79, 0.85] |
| MAE | 4.24 | - | - | 4.67 | 0.60 | [3.87, 6.27] | 3.44 | 0.47 | [2.69, 4.54] |
| RMSE | 9.57 | - | - | 8.41 | 4.96 | [5.78, 22.97] | 7.45 | 2.89 | [5.29, 15.97] |
Note. The forecast performance was evaluated based on participants who consumed alcohol at least once during the study (N = 142). The calculation of ACC, recall, and precision was based on a threshold of 0.3 for both ZIMLP and RS-ZIMLP models. The Mean, SD, and 95% CI represented the means, standard deviations, and 95% credible intervals of the posterior distributions of these measures.