Improve comparison function (#1310)

* Extend data out to 2030-31

* Fix bug in comparison function

* Fix bug in comparison function

Author: nikhilwoodruff

changelog_entry.yaml

@@ -0,0 +1,4 @@
+- bump: patch
+  changes:
+    fixed:
+    - Minor bug in comparison function.

policyengine_uk/microsimulation.py

@@ -80,3 +80,26 @@ def calculate_dataframe(
             return values
         weights = self.get_weights(variable_names[0], period, map_to=map_to)
         return MicroDataFrame(values, weights=weights)
+
+    def compare(
+        self,
+        other: "Simulation",
+        variables: list[str] = None,
+        period: str = None,
+        change_only: bool = False,
+    ):
+        """Compare two simulations for a specific variable list.
+
+        Args:
+            other: Another Simulation instance to compare against
+            variables: List of variable names to compare. If None, compares all variables.
+
+        Returns:
+            DataFrame with comparison results
+        """
+        df = super().compare(
+            other, variables=variables, period=period, change_only=change_only
+        )
+        return MicroDataFrame(
+            df, weights=self.get_weights(variables[0], period)
+        )

policyengine_uk/simulation.py

@@ -539,6 +539,7 @@ def compare(
         other: "Simulation",
         variables: list[str] = None,
         period: str = None,
+        change_only: bool = False,
     ):
         """Compare two simulations for a specific variable list.
 
@@ -569,14 +570,12 @@ def compare(
         for variable in variables:
             self_col = f"{variable}_self"
             other_col = f"{variable}_other"
-            if pd.api.types.is_numeric_dtype(
-                df_combined[self_col]
-            ) and pd.api.types.is_numeric_dtype(df_combined[other_col]):
+            try:
                 df_combined[f"{variable}_change"] = (
                     df_combined[other_col] - df_combined[self_col]
                 )
-            else:
-                # True if different, False if same
+            except:
+                # For other data types, use XOR
                 df_combined[f"{variable}_change"] = (
                     df_combined[other_col] != df_combined[self_col]
                 )
@@ -590,4 +589,17 @@ def compare(
                     f"{variable}_change",
                 ]
             )
+
+        if change_only:
+            variables_to_include = []
+            for variable in variables:
+                if df_combined[f"{variable}_change"].mean() > 0:
+                    variables_to_include.extend(
+                        [
+                            f"{variable}_self",
+                            f"{variable}_other",
+                            f"{variable}_change",
+                        ]
+                    )
+            columns = variables_to_include
         return df_combined[columns]