fix: normalize DataFrame column spaces in predict

pysr/sr.py

@@ -1686,11 +1686,9 @@ def _validate_and_set_fit_params(
                     "Using DataFrame column names instead."
                 )
 
-            if (
-                pd.api.types.is_object_dtype(X.columns)
-                and X.columns.str.contains(" ").any()
-            ):
-                X.columns = X.columns.str.replace(" ", "_")
+            cols_str = X.columns.astype(str)
+            if cols_str.str.contains(" ").any():
+                X.columns = cols_str.str.replace(" ", "_")
                 warnings.warn(
                     "Spaces in DataFrame column names are not supported. "
                     "Spaces have been replaced with underscores. \n"
@@ -2560,6 +2558,19 @@ def predict(
                 # be correctly filtered with self.selection_mask_
                 X = X[X.columns[self.selection_mask_]]
             X.columns = self.feature_names_in_
+
+        # During fit, we replace spaces in DataFrame column names with
+        # underscores. Apply the same normalization here.
+        cols_str = X.columns.astype(str)
+        if cols_str.str.contains(" ").any():
+            X = X.copy()
+            X.columns = cols_str.str.replace(" ", "_")
+            warnings.warn(
+                "Spaces in DataFrame column names are not supported. "
+                "Spaces have been replaced with underscores. \n"
+                "Please rename the columns to valid names."
+            )
+
         # Without feature information, CallableEquation/lambda_format equations
         # require that the column order of X matches that of the X used during
         # the fitting process. _validate_data removes this feature information

pysr/test/test_main.py

@@ -1230,6 +1230,27 @@ def test_pickle_inv_sympy_expression(self):
         # Check that the same operator mapping was used (1/x for inv)
         self.assertEqual(original_result, 0.5 + 3.0)  # 1/2 + 3 = 3.5
 
+    def test_predict_replaces_spaces_in_dataframe_columns(self):
+        # Regression for #690.
+        model = PySRRegressor(
+            niterations=1,
+            populations=1,
+            procs=0,
+            progress=False,
+            verbosity=0,
+            temp_equation_file=True,
+        )
+
+        X_fit = pd.DataFrame({"a b": [1.0, 2.0], "c d": [3.0, 4.0]})
+        y = X_fit["a b"] + X_fit["c d"]
+        X_pred = X_fit.copy()
+
+        model.fit(X_fit, y)
+        np.testing.assert_array_equal(model.feature_names_in_, np.array(["a_b", "c_d"]))
+
+        y_pred = model.predict(X_pred)
+        assert np.isfinite(y_pred).all()
+
     def test_pickle_with_temp_equation_file(self):
         """If we have a temporary equation file, unpickle the estimator."""
         model = PySRRegressor(