fix predict

skfda/ml/regression/_linear_regression.py

@@ -19,7 +19,7 @@
 
 from ..._utils import nquad_vec
 from ..._utils._sklearn_adapter import BaseEstimator, RegressorMixin
-from ...misc._math import inner_product
+from ...misc._math import inner_product_matrix
 from ...misc.lstsq import solve_regularized_weighted_lstsq
 from ...misc.regularization import L2Regularization, compute_penalty_matrix
 from ...representation import FData, FDataBasis
@@ -446,22 +446,25 @@ def predict(  # noqa: D102
         X = self._argcheck_X(X)
         result = []
 
-        if self.functional_response:
+        """ if self.functional_response:
             # Predict each covariate (multivariate or functional)
             if len(X) > 1:
                 return list(
                     itertools.chain.from_iterable(
                         [self.predict(x) for x in X],
                     ),
                 )
-            X = X[0]
+            X = X[0] """
 
         for coef, x, coef_info in zip(self.coef_, X, self._coef_info):
             if self.functional_response:
                 def prediction(arg, x_eval=x, coef_eval=coef):  # noqa: WPS430
                     if isinstance(x_eval, Callable):
                         x_eval = x_eval(arg)  # noqa: WPS220
 
+                    #TODO: MIRAR ESTO BIEN  
+                    x_eval = x_eval.reshape((-1, 1, 1))
+
                     return coef_eval(arg) * x_eval
 
                 result.append(
@@ -470,11 +473,11 @@ def prediction(arg, x_eval=x, coef_eval=coef):  # noqa: WPS430
             else:
                 result.append(coef_info.inner_product(coef, x))
 
-        result = np.sum(result, axis=0)
+        result = sum(result[1:], start=result[0])
 
         if self.fit_intercept:
             if self.functional_response:
-                result[0] = result[0] + self._change_function_basis(
+                result = result + self._change_function_basis(
                     self.intercept_, self.y_basis,
                 )
             else:
@@ -722,14 +725,14 @@ def _change_function_basis(
         if isinstance(f, FDataBasis) and f.basis == new_basis:
             return f
 
-        inner_prod = inner_product(
-            f,
+        inner_prod = inner_product_matrix(
             new_basis,
+            f,
             _domain_range=new_basis.domain_range,
-        )[:, np.newaxis]
+        )
 
         gram_matrix = new_basis.gram_matrix()
 
-        coefs = np.linalg.solve(gram_matrix, inner_prod).flatten()
+        coefs = np.linalg.solve(gram_matrix, inner_prod)
 
-        return FDataBasis(new_basis, coefs)
+        return FDataBasis(new_basis, coefs.T)