diff --git a/lrmodule/copy_csv.py b/lrmodule/copy_csv.py
new file mode 100644
index 0000000..df4939e
--- /dev/null
+++ b/lrmodule/copy_csv.py
@@ -0,0 +1,53 @@
+import shutil
+from pathlib import Path
+
+import pandas as pd
+from lir.aggregation import Aggregation, ContextAwareDict, config_parser, partial, pop_field
+from lir.util import check_type
+
+
+class CopyCSV(Aggregation):
+    """Aggregation that copies a CSV file from a source location to a target location, optionally selecting columns.
+
+    Attributes
+    ----------
+        source_file (str): The path to the source CSV file that should be copied.
+        target_dir (str): The directory where the new CSV file will be saved. Given by the config_parser, meaning
+                            it is not set by the user in the configuration.
+        columns (list[str]): A list of column names to copy from the source CSV. If empty, all columns will be copied.
+        new_file_name (str): The name of the new CSV file. If empty, the original file name will be used.
+    """
+
+    def __init__(self, source_file: str, target_dir: str, columns: list[str], new_file_name: str | None):
+        self.source_file = Path(source_file)
+        self.target_dir = Path(target_dir)
+        self.columns = columns
+        if new_file_name is None:
+            self.new_file_name = self.target_dir / self.source_file.name
+        else:
+            self.new_file_name = self.target_dir / new_file_name
+
+    def report(self, data) -> None:
+        """Do nothing. Required by parent class."""
+        pass
+
+    def close(self):
+        """Close the aggregation and perform any necessary cleanup.
+
+        This method has the logic of this class. It copies the CSV file from the source to the target location,
+        optionally selecting specific columns if they are specified.
+        """
+        if self.columns:
+            df = pd.read_csv(self.source_file)
+            df[self.columns].to_csv(self.new_file_name, index=False)
+        else:
+            shutil.copy(self.source_file, self.new_file_name)
+
+
+@config_parser()
+def copy_csv(config: ContextAwareDict, output_dir: str) -> CopyCSV:
+    """Parse the configuration for the CopyCSV aggregation and return an instance of it."""
+    source_file = pop_field(config, "file")
+    columns = pop_field(config, "columns", default=[], validate=partial(check_type, list))
+    new_file_name = pop_field(config, "new_file_name", required=False)
+    return CopyCSV(source_file, output_dir, columns=columns, new_file_name=new_file_name)
diff --git a/validation.yaml b/validation.yaml
index 48a07a2..7f88546 100644
--- a/validation.yaml
+++ b/validation.yaml
@@ -186,6 +186,11 @@ experiments:
       - pav
       - ece
       - llr_interval
+      - method: lrmodule.copy_csv.copy_csv
+        file: ${data_root}/breech_face_impression.csv
+        new_file_name: source_data.csv
+      - save_model
+
 
   - name: breech_face_impression_model_cmc
     strategy: single_run