Tip: use “Explain prediction” while choose a candidate.
Metrics
Metric
Value
Accuracy
Precision
Recall
ROC-AUC
Confusion Matrix
Feature importance / SHAP
If SHAP isn’t available from backend, θα δεις fallback feature importance.
Dataset format (CSV)
Required: a header row with at least label and one numeric feature column.
Optional identifiers: target, tic κ.λπ. Example header:target,label,period,duration,depth,power,snr,delta_bic
Labels: 0 = non-planet, 1 = planet.
Advanced: Upload custom light curve (TXT/CSV)
Upload TXT (2 columns: time, flux)
Light Curve
Phase-Folded (selected candidate)
Candidates (green = vetted with P ≥ threshold; badge = centroid test)