bts classifier

src/pfaedle/PfaedleMain.cpp

@@ -61,6 +61,10 @@ using pfaedle::router::RouterImpl;
 using pfaedle::router::ShapeBuilder;
 using pfaedle::router::Stats;
 using pfaedle::statsimiclassifier::JaccardClassifier;
+using pfaedle::statsimiclassifier::StatsimiClassifier;
+using pfaedle::statsimiclassifier::BTSClassifier;
+using pfaedle::statsimiclassifier::EDClassifier;
+using pfaedle::statsimiclassifier::PEDClassifier;
 
 enum class RetCode {
   SUCCESS = 0,
@@ -277,7 +281,21 @@ int main(int argc, char** argv) {
         graphDimensions[filePost].second += nd->getAdjListOut().size();
       }
 
-      JaccardClassifier statsimiClassifier;
+      StatsimiClassifier* statsimiClassifier;
+
+      if (motCfg.routingOpts.statsimiMethod == "bts") {
+        statsimiClassifier = new BTSClassifier();
+      } else if (motCfg.routingOpts.statsimiMethod == "jaccard") {
+        statsimiClassifier = new JaccardClassifier();
+      } else if (motCfg.routingOpts.statsimiMethod == "ed") {
+        statsimiClassifier = new EDClassifier();
+      } else if (motCfg.routingOpts.statsimiMethod == "ped") {
+        statsimiClassifier = new PEDClassifier();
+      } else {
+        LOG(ERROR) << "Unknown station similarity classifier "
+                   << motCfg.routingOpts.statsimiMethod;
+        exit(1);
+      }
 
       Router* router = 0;
 
@@ -303,7 +321,7 @@ int main(int argc, char** argv) {
       }
 
       ShapeBuilder shapeBuilder(&gtfs[0], usedMots, motCfg, &graph, &fStops,
-                                &restr, &statsimiClassifier, router, cfg);
+                                &restr, statsimiClassifier, router, cfg);
 
       pfaedle::netgraph::Graph ng;
 
@@ -326,6 +344,7 @@ int main(int argc, char** argv) {
       }
 
       if (router) delete router;
+      if (statsimiClassifier) delete statsimiClassifier;
 
       if (cfg.writeGraph) {
         LOG(INFO) << "Outputting graph.json...";

src/pfaedle/config/MotConfigReader.cpp

@@ -61,6 +61,13 @@ void MotConfigReader::parse(const std::vector<std::string>& paths,
       cfg.routingOpts.transPenMethod = "exp";
     }
 
+    if (p.hasKey(secStr, "station_similarity_classification_method")) {
+      cfg.routingOpts.statsimiMethod =
+          p.getStr(secStr, "station_similarity_classification_method");
+    } else {
+      cfg.routingOpts.statsimiMethod = "bts";
+    }
+
     if (p.hasKey(secStr, "routing_use_stations")) {
       cfg.routingOpts.useStations = p.getBool(secStr, "routing_use_stations");
     } else {

src/pfaedle/router/Misc.h

@@ -62,6 +62,7 @@ struct RoutingOpts {
   double transitionPen;
   std::string transPenMethod;
   std::string emPenMethod;
+  std::string statsimiMethod;
 };
 
 // _____________________________________________________________________________
@@ -82,6 +83,7 @@ inline bool operator==(const RoutingOpts& a, const RoutingOpts& b) {
          fabs(a.nonStationPen - b.nonStationPen) < 0.01 &&
          a.transPenMethod == b.transPenMethod &&
          a.emPenMethod == b.emPenMethod &&
+         a.statsimiMethod == b.statsimiMethod &&
          a.useStations == b.useStations && a.popReachEdge == b.popReachEdge &&
          a.noSelfHops == b.noSelfHops;
 }

src/pfaedle/router/ShapeBuilder.cpp

@@ -47,7 +47,6 @@ using pfaedle::router::ShapeBuilder;
 using pfaedle::router::Stats;
 using pfaedle::router::TripForests;
 using pfaedle::router::TripTrie;
-using pfaedle::statsimiclassifier::JaccardClassifier;
 using pfaedle::trgraph::EdgeGrid;
 using pfaedle::trgraph::NodeGrid;
 using util::geo::latLngToWebMerc;

src/pfaedle/statsimi-classifier/StatsimiClassifier.cpp

@@ -9,7 +9,10 @@
 #include "pfaedle/statsimi-classifier/StatsimiClassifier.h"
 #include "util/geo/Geo.h"
 
+using pfaedle::statsimiclassifier::BTSClassifier;
+using pfaedle::statsimiclassifier::EDClassifier;
 using pfaedle::statsimiclassifier::JaccardClassifier;
+using pfaedle::statsimiclassifier::PEDClassifier;
 
 // _____________________________________________________________________________
 bool JaccardClassifier::similar(const std::string& nameA, const POINT& posA,
@@ -23,8 +26,52 @@ bool JaccardClassifier::similar(const std::string& nameA, const POINT& posA,
 // _____________________________________________________________________________
 bool JaccardClassifier::similar(const std::string& nameA,
                                 const std::string& nameB) const {
-  // hard similarity
-  if (nameA == nameB) return true;
-
-  return util::jaccardSimi(nameA, nameB) > 0.45;  // 0.45 from paper
+  return util::jaccardSimi(nameA, nameB) > 0.45;  // 0.45 from statsimi paper
+}
+
+// _____________________________________________________________________________
+bool BTSClassifier::similar(const std::string& nameA, const POINT& posA,
+                            const std::string& nameB, const POINT& posB) const {
+  UNUSED(posA);
+  UNUSED(posB);
+  return similar(nameA, nameB);
+}
+
+// _____________________________________________________________________________
+bool BTSClassifier::similar(const std::string& nameA,
+                            const std::string& nameB) const {
+  return util::btsSimi(nameA, nameB) > 0.85;  // 0.85 from statsimi paper
+}
+
+// _____________________________________________________________________________
+bool EDClassifier::similar(const std::string& nameA, const POINT& posA,
+                           const std::string& nameB, const POINT& posB) const {
+  UNUSED(posA);
+  UNUSED(posB);
+  return similar(nameA, nameB);
+}
+
+// _____________________________________________________________________________
+bool EDClassifier::similar(const std::string& nameA,
+                           const std::string& nameB) const {
+  double edSimi = 1.0 - ((util::editDist(nameA, nameB) * 1.0) /
+                         fmax(nameA.size(), nameB.size()));
+  return edSimi > 0.85;  // 0.85 from statsimi paper
+}
+
+// _____________________________________________________________________________
+bool PEDClassifier::similar(const std::string& nameA, const POINT& posA,
+                            const std::string& nameB, const POINT& posB) const {
+  UNUSED(posA);
+  UNUSED(posB);
+  return similar(nameA, nameB);
+}
+
+// _____________________________________________________________________________
+bool PEDClassifier::similar(const std::string& nameA,
+                            const std::string& nameB) const {
+  double a = (util::prefixEditDist(nameA, nameB) * 1.0) / (nameA.size() * 1.0);
+  double b = (util::prefixEditDist(nameB, nameA) * 1.0) / (nameB.size() * 1.0);
+  double pedSimi = 1.0 - fmin(a, b);
+  return pedSimi > 0.875;  // 0.875 average of values from statsimi paper
 }

src/pfaedle/statsimi-classifier/StatsimiClassifier.h

@@ -14,6 +14,7 @@ namespace statsimiclassifier {
 
 class StatsimiClassifier {
  public:
+  virtual ~StatsimiClassifier() {}
   virtual bool similar(const std::string& nameA, const POINT& posA,
                        const std::string& nameB, const POINT& posB) const = 0;
 
@@ -29,6 +30,30 @@ class JaccardClassifier : public StatsimiClassifier {
                        const std::string& nameB) const;
 };
 
+class BTSClassifier : public StatsimiClassifier {
+ public:
+  virtual bool similar(const std::string& nameA, const POINT& posA,
+                       const std::string& nameB, const POINT& posB) const;
+  virtual bool similar(const std::string& nameA,
+                       const std::string& nameB) const;
+};
+
+class EDClassifier : public StatsimiClassifier {
+ public:
+  virtual bool similar(const std::string& nameA, const POINT& posA,
+                       const std::string& nameB, const POINT& posB) const;
+  virtual bool similar(const std::string& nameA,
+                       const std::string& nameB) const;
+};
+
+class PEDClassifier : public StatsimiClassifier {
+ public:
+  virtual bool similar(const std::string& nameA, const POINT& posA,
+                       const std::string& nameB, const POINT& posB) const;
+  virtual bool similar(const std::string& nameA,
+                       const std::string& nameB) const;
+};
+
 }  // namespace statsimiclassifier
 }  // namespace pfaedle