make jaccard-geodist the default statsimi classifier

pfaedle.cfg

@@ -7,7 +7,7 @@
 routing_transition_penalty_fac:     0.0083
 routing_station_move_penalty_fac:	0.002
 
-station_similarity_classification_method: jaccard
+station_similarity_classification_method: jaccard-geodist
 
 # Regular expressions and station comparision is
 # always case insensitive!

src/pfaedle/PfaedleMain.cpp

@@ -61,6 +61,7 @@ using pfaedle::router::RouterImpl;
 using pfaedle::router::ShapeBuilder;
 using pfaedle::router::Stats;
 using pfaedle::statsimiclassifier::JaccardClassifier;
+using pfaedle::statsimiclassifier::JaccardGeodistClassifier;
 using pfaedle::statsimiclassifier::StatsimiClassifier;
 using pfaedle::statsimiclassifier::BTSClassifier;
 using pfaedle::statsimiclassifier::EDClassifier;
@@ -288,6 +289,8 @@ int main(int argc, char** argv) {
         statsimiClassifier = new BTSClassifier();
       } else if (motCfg.routingOpts.statsimiMethod == "jaccard") {
         statsimiClassifier = new JaccardClassifier();
+      } else if (motCfg.routingOpts.statsimiMethod == "jaccard-geodist") {
+        statsimiClassifier = new JaccardGeodistClassifier();
       } else if (motCfg.routingOpts.statsimiMethod == "ed") {
         statsimiClassifier = new EDClassifier();
       } else if (motCfg.routingOpts.statsimiMethod == "ped") {

src/pfaedle/config/MotConfigReader.cpp

@@ -65,7 +65,7 @@ void MotConfigReader::parse(const std::vector<std::string>& paths,
       cfg.routingOpts.statsimiMethod =
           p.getStr(secStr, "station_similarity_classification_method");
     } else {
-      cfg.routingOpts.statsimiMethod = "bts";
+      cfg.routingOpts.statsimiMethod = "jaccard-geodist";
     }
 
     if (p.hasKey(secStr, "routing_use_stations")) {

src/pfaedle/statsimi-classifier/StatsimiClassifier.cpp

@@ -12,8 +12,35 @@
 using pfaedle::statsimiclassifier::BTSClassifier;
 using pfaedle::statsimiclassifier::EDClassifier;
 using pfaedle::statsimiclassifier::JaccardClassifier;
+using pfaedle::statsimiclassifier::JaccardGeodistClassifier;
 using pfaedle::statsimiclassifier::PEDClassifier;
 
+// _____________________________________________________________________________
+bool JaccardGeodistClassifier::similar(const std::string& nameA,
+                                       const POINT& posA,
+                                       const std::string& nameB,
+                                       const POINT& posB) const {
+  const double THRES_M =
+      0.00815467271246994481;    // ln 2/85 from statsimi evaluation
+  const double THRES_JACC = .5;  // from statsimi evaluation
+
+  const double m = exp(-THRES_M * util::geo::haversine(posA, posB));
+  double jacc = util::jaccardSimi(nameA, nameB);
+
+  if (jacc > THRES_JACC)
+    jacc = .5 + (jacc - THRES_JACC) / (2.0 * (1.0 - THRES_JACC));
+  else
+    jacc = jacc / (2.0 * THRES_JACC);
+
+  return (m + jacc) / 2.0;
+}
+
+// _____________________________________________________________________________
+bool JaccardGeodistClassifier::similar(const std::string& nameA,
+                                       const std::string& nameB) const {
+  return util::jaccardSimi(nameA, nameB) > 0.45;  // 0.45 from statsimi paper
+}
+
 // _____________________________________________________________________________
 bool JaccardClassifier::similar(const std::string& nameA, const POINT& posA,
                                 const std::string& nameB,

src/pfaedle/statsimi-classifier/StatsimiClassifier.h

@@ -30,6 +30,14 @@ class JaccardClassifier : public StatsimiClassifier {
                        const std::string& nameB) const;
 };
 
+class JaccardGeodistClassifier : 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 BTSClassifier : public StatsimiClassifier {
  public:
   virtual bool similar(const std::string& nameA, const POINT& posA,