subset = ["start_lng", "start_lat", "start_station_id"]
all_starts = trips.drop_duplicates(subset="start_station_id", keep="first")[subset]

subset = ["end_lng", "end_lat", "end_station_id"]
all_ends = trips.drop_duplicates(subset="end_station_id", keep="first")[subset]
distance_matrix = all_ends.merge(all_starts, how="cross")
distance_matrix["distance"] = distance_matrix.apply(lambda x: geodesic((x["start_lat"], x["start_lng"]), 
                                                        (x["end_lat"], x["end_lng"])).meters, axis=1)
#considering a specific treshhold
distance_matrix["edge"] = distance_matrix["distance"] < 500