more work on model training

scripts/run_p0_rain_workflow.sh

@@ -8,6 +8,9 @@ MODEL_VERSION="${MODEL_VERSION:-rain-logreg-v1}"
 MODEL_PATH="${MODEL_PATH:-models/rain_model.pkl}"
 REPORT_PATH="${REPORT_PATH:-models/rain_model_report.json}"
 AUDIT_PATH="${AUDIT_PATH:-models/rain_data_audit.json}"
+FEATURE_SET="${FEATURE_SET:-baseline}"
+FORECAST_MODEL="${FORECAST_MODEL:-ecmwf}"
+DATASET_PATH="${DATASET_PATH:-models/datasets/rain_dataset_${MODEL_VERSION}_${FEATURE_SET}.csv}"
 
 if [[ -z "${DATABASE_URL:-}" ]]; then
   echo "DATABASE_URL is required"
@@ -20,6 +23,8 @@ python scripts/audit_rain_data.py \
   --site "$SITE" \
   --start "$START" \
   --end "$END" \
+  --feature-set "$FEATURE_SET" \
+  --forecast-model "$FORECAST_MODEL" \
   --out "$AUDIT_PATH"
 
 echo "Training baseline rain model..."
@@ -30,14 +35,18 @@ python scripts/train_rain_model.py \
   --train-ratio 0.7 \
   --val-ratio 0.15 \
   --min-precision 0.70 \
+  --feature-set "$FEATURE_SET" \
+  --forecast-model "$FORECAST_MODEL" \
   --model-version "$MODEL_VERSION" \
   --out "$MODEL_PATH" \
-  --report-out "$REPORT_PATH"
+  --report-out "$REPORT_PATH" \
+  --dataset-out "$DATASET_PATH"
 
 echo "Writing current prediction..."
 python scripts/predict_rain_model.py \
   --site "$SITE" \
   --model-path "$MODEL_PATH" \
-  --model-name "rain_next_1h"
+  --model-name "rain_next_1h" \
+  --forecast-model "$FORECAST_MODEL"
 
 echo "P0 rain workflow complete."