fix(CLI): Fix model scoring in CLI functions

Author: nmaarnio

eis_toolkit/cli.py

@@ -2318,20 +2318,23 @@ def evaluate_trained_model_cli(
     output_raster: OUTPUT_FILE_OPTION,
     validation_metrics: Annotated[List[str], typer.Option()],
 ):
-    """Train and optionally validate a Gradient boosting regressor model using Sklearn."""
-    from eis_toolkit.prediction.machine_learning_general import (
-        evaluate_model,
-        load_model,
-        prepare_data_for_ml,
-        reshape_predictions,
-    )
+    """Evaluate a trained machine learning model by predicting and scoring."""
+    from sklearn.base import is_classifier
 
-    X, y, reference_profile, nodata_mask = prepare_data_for_ml(input_rasters, target_labels)
+    from eis_toolkit.evaluation.scoring import score_predictions
+    from eis_toolkit.prediction.machine_learning_general import load_model, prepare_data_for_ml, reshape_predictions
+    from eis_toolkit.prediction.machine_learning_predict import predict_classifier, predict_regressor
 
+    X, y, reference_profile, nodata_mask = prepare_data_for_ml(input_rasters, target_labels)
+    print(len(np.unique(y)))
     typer.echo("Progress: 30%")
 
     model = load_model(model_file)
-    predictions, metrics_dict = evaluate_model(X, y, model, validation_metrics)
+    if is_classifier(model):
+        predictions, probabilities = predict_classifier(X, model, True)
+    else:
+        predictions = predict_regressor(X, model)
+    metrics_dict = score_predictions(y, predictions, validation_metrics)
     predictions_reshaped = reshape_predictions(
         predictions, reference_profile["height"], reference_profile["width"], nodata_mask
     )
@@ -2359,20 +2362,22 @@ def predict_with_trained_model_cli(
     model_file: INPUT_FILE_OPTION,
     output_raster: OUTPUT_FILE_OPTION,
 ):
-    """Train and optionally validate a Gradient boosting regressor model using Sklearn."""
-    from eis_toolkit.prediction.machine_learning_general import (
-        load_model,
-        predict,
-        prepare_data_for_ml,
-        reshape_predictions,
-    )
+    """Predict with a trained machine learning model."""
+    from sklearn.base import is_classifier
+
+    from eis_toolkit.prediction.machine_learning_general import load_model, prepare_data_for_ml, reshape_predictions
+    from eis_toolkit.prediction.machine_learning_predict import predict_classifier, predict_regressor
 
     X, _, reference_profile, nodata_mask = prepare_data_for_ml(input_rasters)
 
     typer.echo("Progress: 30%")
 
     model = load_model(model_file)
-    predictions = predict(X, model)
+    if is_classifier(model):
+        predictions, probabilities = predict_classifier(X, model, True)
+    else:
+        predictions = predict_regressor(X, model)
+
     predictions_reshaped = reshape_predictions(
         predictions, reference_profile["height"], reference_profile["width"], nodata_mask
     )