How a notebook model became a tested, containerized, monitored inference service. Every figure on this page comes from the project repository.
| Class | Precision | Recall | F1 | Orders |
|---|---|---|---|---|
| On time | 0.94 | 0.99 | 0.96 | 13,515 |
| Delayed | 0.31 | 0.08 | 0.13 | 957 |
| Macro average | 0.63 | 0.54 | 0.55 | 14,472 |
POST /predictone orderPOST /predict/batchmany orders in one callGET /healthliveness checkGET /model-infowhich model version is servingGET /monitoring/summarydrift reportGET /metricsPrometheus scrape target