Fixes #301 and some other issues

pygam/pygam.py

@@ -386,7 +386,7 @@ def _linear_predictor(self, X=None, modelmat=None, b=None, term=-1):
         ---------
         at least 1 of (X, modelmat)
             and
-        at least 1 of (b, feature)
+        at least 1 of (b, term)
 
         X : array-like of shape (n_samples, m_features) or None, optional
             containing the input dataset
@@ -401,9 +401,9 @@ def _linear_predictor(self, X=None, modelmat=None, b=None, term=-1):
             contains the spline coefficients
             if None, will use current model coefficients
 
-        feature : int, optional
-                  feature for which to compute the linear prediction
-                  if -1, will compute for all features
+        term : int or list of int, default: -1
+            term(s) to use in calculation of linear predictor
+            if -1, all terms are used
 
         Returns
         -------
@@ -480,12 +480,22 @@ def _modelmat(self, X, term=-1):
         modelmat : sparse matrix of len n_samples
             containing model matrix of the spline basis for selected features
         """
+        # take features, dtypes and edge_knots based on supplied term values
+        if term != -1:
+            terms = list(np.atleast_1d(term))
+        else:
+            terms = range(len(self.feature))
+
+        features = [self.feature[i] for i in terms]
+        edge_knots = [self.edge_knots_[i] for i in terms]
+        dtypes = [self.dtype[i] for i in terms]
+
         X = check_X(
             X,
             n_feats=self.statistics_['m_features'],
-            edge_knots=self.edge_knots_,
-            dtypes=self.dtype,
-            features=self.feature,
+            edge_knots=edge_knots,
+            dtypes=dtypes,
+            features=features,
             verbose=self.verbose,
         )
 
@@ -1434,6 +1444,9 @@ def _flatten_mesh(self, Xs, term):
         X = np.zeros((n, self.statistics_['m_features']))
         for term_, x in zip(terms, Xs):
             X[:, term_.feature] = x.ravel()
+
+        if getattr(self.terms[term], 'by', None) is not None:
+            X[:, self.terms[term].by] = 1.0
         return X
 
     def generate_X_grid(self, term, n=100, meshgrid=False):
@@ -1607,14 +1620,6 @@ def partial_dependence(
             shape = X[0].shape
 
             X = self._flatten_mesh(X, term=term)
-            X = check_X(
-                X,
-                n_feats=self.statistics_['m_features'],
-                edge_knots=self.edge_knots_,
-                dtypes=self.dtype,
-                features=self.feature,
-                verbose=self.verbose,
-            )
 
         modelmat = self._modelmat(X, term=term)
         pdep = self._linear_predictor(modelmat=modelmat, term=term)

pygam/terms.py

@@ -846,7 +846,6 @@ def build_columns(self, X, verbose=False):
         -------
         scipy sparse array with n rows
         """
-        X[:, self.feature][:, np.newaxis]
 
         splines = b_spline_basis(
             X[:, self.feature],