diff --git a/app/preprocessing.py b/app/preprocessing.py
index 400473c..c276843 100644
--- a/app/preprocessing.py
+++ b/app/preprocessing.py
@@ -18,4 +18,5 @@ def create_dataframe(dataset) -> DataFrame:
     """
     json = open_json(dataset)
     df = json_normalize(data=json, record_path=["records"], errors="ignore",)
-    return df
+    filtered_df = df.filter(items=COLUMNS[dataset])
+    return filtered_df