| Research History |
| Software Structure |
| Specie Sensitivity Distribution |
| BAYESIAN Inference |
| MCMC Simulation |
| DIC Optimization |
| Ecorisk & Uncertainty |
| Joint Probability Curve |
| Exergy SSD |
| Main Function Lists Panel |
| BMC-SSD Panel |
| Models Optimiztion Panel |
| JPC Panel |
| ExSSD Panel |
| Work Path & Output Results |
| Installation & Initialization |
| Folder & File Extraction |
| SSD Models & Ecorisk |
| JPC & Its Indicators |
| Models Optimization & Parameters |
| ExSSD Models & ExEcorisk |
Links
| College of Urban and Environment Science |
| Peking University |
DIC Optimization
Normally, optimization model is defined as a relatively simple model which covers relatively more information. Spiegelhalter et al. (2002) proposed the deviance information criterion (DIC, a criterion like Akaike Information Criterion) to optimize the model. DIC can be expressed as: DIC = D + 2pD. where D denotes the fitting degree and 2pD evaluates simplicity of the model. Both D and 2pD can be calculated by MCMC simulation. Generally, models with lower DIC values tend to be better. However, according to Spiegelhalter et al. (2002)’s DIC difference standard between models (ΔDIC), when ΔDIC < 2, the two models are considered equally good, when ΔDIC = 4~7, the two models can be considered as good as each other but the extent of equality is weak, and when ΔDIC > 10, the model with lower DIC is better. During actual operation, models are more likely to be considered as equally good, therefore, in our software, D is chosen as the second standard—the SSD model with lower D value is selected as the best among the several optimization models selected by the ΔDIC rule.
