Machine-learning intelligence
Insights Lab
Pull the latest predictive outputs and turn them into actionable scouting signals.
Visual intelligence layer
Models, charts, and pipeline at a glance
React Flow shows the conceptual training path; Recharts panels light up when JSON artifacts are mounted in the API container. Scroll down for filtered embryo lists, CSV export, and tactical notes — this section is the “generous” overview.
Artifacts online
4/4
Price, enriched SHAP, advanced stack, embryo run.
Best price CV R²
0.164
XGBoost
Embryos scored
318
483 in last export
Top SHAP features
6
Mean |SHAP| ranking from enriched/price report.
ML pipeline (conceptual)
Same logical path as enrichment/price_predictor.py and advanced_models.py
Model comparison (cross-validated R²)
Verdict mix (embryo run)
Confidence distribution
Global feature influence (mean |SHAP|)
How to read this lab
- The flow is a conceptual mirror of the Python training stack — not a live DAG of Airflow jobs.
- R² bars compare models on log-price or price scale depending on the exported JSON; use them for relative ranking, not FEI rulemaking.
- Embryo charts only light up when
embryo_predictions.jsonis mounted; price/SHAP panels needprice_model_report.json.