Predictive performance by temporal period under balanced sampling
| Train year | Test year | AUC | Precision | Recall | F1-Score |
|---|---|---|---|---|---|
| 2018 | 2019 | 0.972 | 0.899 | 0.625 | 0.716 |
| 2019 | 2020 | 0.968 | 0.933 | 0.622 | 0.723 |
| 2020 | 2021 | 0.957 | 0.896 | 0.531 | 0.640 |
| 2021 | 2022 | 0.960 | 0.919 | 0.502 | 0.636 |
| Train year | Test year | Precision | Recall | F1-Score | |
|---|---|---|---|---|---|
| 2018 | 2019 | 0.972 | 0.899 | 0.625 | 0.716 |
| 2019 | 2020 | 0.968 | 0.933 | 0.622 | 0.723 |
| 2020 | 2021 | 0.957 | 0.896 | 0.531 | 0.640 |
| 2021 | 2022 | 0.960 | 0.919 | 0.502 | 0.636 |
Values represent averages across 12 CPV categories for each temporal period. The narrow range in AUC (0.957–0.972) and precision (0.896–0.933) demonstrates temporal stability of structural prediction patterns, while modest variation in recall reflects either changing procurement dynamics or increased relationship complexity in later years
Sharing content requires targeting cookies to be enabled. Please update your cookie preferences to use this feature.