expose mapping_distance

libpysal/graph/_network.py

@@ -61,11 +61,7 @@ def pdna_to_adj(origins, network, node_ids, threshold):
     return adj
 
 
-def build_travel_graph(
-    df,
-    network,
-    threshold,
-):
+def build_travel_graph(df, network, threshold, mapping_distance):
     """Compute the shortest path between gdf centroids via a pandana.Network
     and return an adjacency list with weight=cost. Note unlike distance_band,
     :math:`G_{ij}` and :math:`G_{ji}` are often different because travel networks
@@ -81,6 +77,10 @@ def build_travel_graph(
         Network that encodes travel costs. See <https://udst.github.io/pandana/>
     threshold : int
         maximum travel cost to consider neighbors
+    mapping_distance : int
+        snapping tolerance passed to ``pandana.Network.get_node_ids`` that defines
+        the maximum range at which observations are snapped to nearest nodes in the
+        network. Default is None
 
     Returns
     -------
@@ -89,7 +89,9 @@ def build_travel_graph(
     """
     df = df.copy()
     _validate_geometry_input(df.geometry, ids=None, valid_geometry_types="Point")
-    df["node_ids"] = network.get_node_ids(df.geometry.x, df.geometry.y)
+    df["node_ids"] = network.get_node_ids(
+        df.geometry.x, df.geometry.y, mapping_distance
+    )
 
     # depending on density of the graph nodes / observations, it is common to have
     # multiple observations snapped to the same network node, so use the clique

libpysal/graph/base.py

@@ -1456,18 +1456,21 @@ def build_h3(cls, ids, order=1, weight="distance"):
             raise ValueError("weight must be one of 'distance', 'binary', or 'inverse'")
 
     @classmethod
-    def build_travel_cost(cls, df, network, threshold, kernel=None):
+    def build_travel_cost(
+        cls, df, network, threshold, kernel=None, mapping_distance=None
+    ):
         """Generate a Graph based on shortest travel costs from a pandana.Network
 
         Parameters
         ----------
         df : geopandas.GeoDataFrame
             geodataframe representing observations which are snapped to the nearest
-            node in the pandana.Network. If passing polygon geometries, the spatial
-            support will be reduced to Points (via centroid) before snapping.
+            node in the pandana.Network. CRS should be the same as the locations
+            of ``node_x`` and ``node_y`` in the pandana.Network (usually 4326 if network
+            comes from OSM, but sometimes projected to improve snapping quality).
         network : pandana.Network
             pandana Network object describing travel costs between nodes in the study
-            area
+            area.  See <https://udst.github.io/pandana/> for more
         threshold : int
             threshold representing maximum cost distances. This is measured in the same
             units as the pandana.Network (not influenced by the df.crs in any way). For
@@ -1480,6 +1483,10 @@ def build_travel_cost(cls, df, network, threshold, kernel=None):
             libpysal.graph.Graph.build_kernel for more information on kernel
             transformation options. Default is None, in which case the Graph weight
             is pure distance between focal and neighbor
+        mapping_distance : int
+            snapping tolerance passed to ``pandana.Network.get_node_ids`` that defines
+            the maximum range at which observations are snapped to nearest nodes in the
+            network. Default is None
 
         Returns
         -------
@@ -1524,7 +1531,7 @@ def build_travel_cost(cls, df, network, threshold, kernel=None):
                 117         333.639008
         Name: weight, dtype: float64
         """
-        adj = _build_travel_graph(df, network, threshold)
+        adj = _build_travel_graph(df, network, threshold, mapping_distance)
         g = cls.from_adjacency(adj)
         if kernel is not None:
             arrays = _kernel(