4 files changed, 139 insertions, 42 deletions

diff --git a/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/CostDistance.xtend b/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/CostDistance.xtend
index 33d10fa3..613f0f43 100644
--- a/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/CostDistance.xtend
+++ b/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/CostDistance.xtend
@@ -1,9 +1,25 @@
 package ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator.distance
 
+import java.text.DecimalFormat
+import java.util.HashMap
+import java.util.List
 import org.eclipse.xtend.lib.annotations.Accessors
 
-class CostDistance {
-
+abstract class CostDistance {
+        def abstract double naDistance(List<Double> samples);   
+        def abstract double mpcDistance(List<Double> samples);
+        def abstract double outDegreeDistance(List<Double> samples);
+        
+        def protected pmfFromSamples(double[] samples, DecimalFormat formatter){
+                var length = samples.length;
+                var pmfMap = new HashMap<String, Double>();
+                
+                for(sample : samples){
+                        pmfMap.put(formatter.format(sample), pmfMap.getOrDefault(formatter.format(sample), 0.0) + 1.0 / length);
+                }
+                
+                return pmfMap;
+        }
 }
 
 class StateData{
diff --git a/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/EuclideanDistance.xtend b/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/EuclideanDistance.xtend
new file mode 100644
index 00000000..b945d97b
--- /dev/null
+++ b/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/EuclideanDistance.xtend
@@ -0,0 +1,73 @@
+package ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator.distance
+
+import ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator.app.Domain
+import ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator.io.RepMetricsReader
+import ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator.metrics.MetricSampleGroup
+import com.google.common.collect.Sets
+import java.text.DecimalFormat
+import java.util.ArrayList
+import java.util.HashMap
+import java.util.List
+import java.util.Map
+import java.util.Set
+
+class EuclideanDistance extends CostDistance{
+        var MetricSampleGroup g;
+        var HashMap<String, Double> mpcPMF;
+        var HashMap<String, Double> naPMF;
+        var HashMap<String, Double> outDegreePMF;
+        var DecimalFormat formatter;
+        
+        new(Domain d){
+                var metrics = RepMetricsReader.read(d);
+                this.g = metrics;
+                
+                var mpcSamples = metrics.mpcSamples;
+                var naSamples = metrics.naSamples.stream.mapToDouble([it]).toArray();
+                var outDegreeSamples = metrics.outDegreeSamples.stream.mapToDouble([it]).toArray();
+                
+                //needs to format the number to string avoid precision issue
+                formatter = new DecimalFormat("#0.00000");  
+                
+                mpcPMF = pmfFromSamples(mpcSamples, formatter);
+                naPMF = pmfFromSamples(naSamples, formatter);
+                outDegreePMF = pmfFromSamples(outDegreeSamples, formatter);     
+        }
+        
+        override naDistance(List<Double> samples) {
+                var pmfMap = pmfFromSamples(samples, formatter);
+                return euclideanDistance(pmfMap, naPMF);
+        }
+        
+        override mpcDistance(List<Double> samples) {
+                var pmfMap = pmfFromSamples(samples, formatter);
+                return euclideanDistance(pmfMap, mpcPMF);
+        }
+        
+        override outDegreeDistance(List<Double> samples) {
+                var pmfMap = pmfFromSamples(samples, formatter);
+                return euclideanDistance(pmfMap, outDegreePMF);
+        }
+        
+        
+        def private euclideanDistance(HashMap<String, Double> pmf1, HashMap<String, Double> pmf2){
+                var keys = Sets.union(pmf1.keySet(), pmf2.keySet());
+                var pmfList1 = pmfMapToList(pmf1, keys);
+                var pmfList2 = pmfMapToList(pmf2, keys);
+                var distance = 0.0;
+                for(var i = 0; i < pmfList1.size(); i++){
+                        distance += Math.pow(pmfList1.get(i) + pmfList2.get(i), 2);
+                }
+                
+                return Math.sqrt(distance);
+        }       
+        
+        def private pmfMapToList(Map<String, Double> map, Set<String> keys){
+                var list = new ArrayList<Double>();
+                for(key : keys){
+                        var value = map.getOrDefault(key, 0.0);
+                        list.add(value);
+                }
+                return list;
+        }
+}
\ No newline at end of file
diff --git a/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/JSDistance.xtend b/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/JSDistance.xtend
index ced9eadb..df65b81f 100644
--- a/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/JSDistance.xtend
+++ b/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/JSDistance.xtend
@@ -7,7 +7,7 @@ import java.text.DecimalFormat
 import java.util.HashMap
 import java.util.List
 
-class JSDistance {
+class JSDistance extends CostDistance {
         var HashMap<String, Double> mpcPMF;
         var HashMap<String, Double> naPMF;
         var HashMap<String, Double> outDegreePMF;
@@ -27,17 +27,6 @@ class JSDistance {
                 outDegreePMF = pmfFromSamples(outDegreeSamples, formatter);     
         }
         
-        def private pmfFromSamples(double[] samples, DecimalFormat formatter){
-                var length = samples.length;
-                var pmfMap = new HashMap<String, Double>();
-                
-                for(sample : samples){
-                        pmfMap.put(formatter.format(sample), pmfMap.getOrDefault(formatter.format(sample), 0.0) + 1.0 / length);
-                }
-                
-                return pmfMap;
-        }
-        
         def private combinePMF(HashMap<String, Double> pmf1, HashMap<String, Double> pmf2){
                 var pmfMap = new HashMap<String, Double>();
                 
@@ -68,7 +57,7 @@ class JSDistance {
                 return distance;
         }
         
-        def double mpcDistance(List<Double> samples){
+        override double mpcDistance(List<Double> samples){
                 // map list to array
                 var map = pmfFromSamples(samples.stream().mapToDouble([it]).toArray(), formatter);
                 //if the size of array is smaller than 2, ks distance cannot be performed, simply return 1
@@ -76,7 +65,7 @@ class JSDistance {
                 return jsDivergence(map, mpcPMF);
         }
         
-        def double naDistance(List<Double> samples){
+        override double naDistance(List<Double> samples){
                 // map list to array
                 var map = pmfFromSamples(samples.stream().mapToDouble([it]).toArray(), formatter);
                 
@@ -85,7 +74,7 @@ class JSDistance {
                 return jsDivergence(map, naPMF);
         }
         
-        def double outDegreeDistance(List<Double> samples){
+        override double outDegreeDistance(List<Double> samples){
                 // map list to array
                 var map = pmfFromSamples(samples.stream().mapToDouble([it]).toArray(), formatter);
                 //if the size of array is smaller than 2, ks distance cannot be performed, simply return 1
diff --git a/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/KSDistance.xtend b/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/KSDistance.xtend
index 58e0a8a3..86f5f23c 100644
--- a/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/KSDistance.xtend
+++ b/Metrics/Metrics-Calculation/ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator/src/ca/mcgill/ecse/dslreasoner/realistic/metrics/calculator/distance/KSDistance.xtend
@@ -2,46 +2,65 @@ package ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator.distance
 
 import ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator.app.Domain
 import ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator.io.RepMetricsReader
+import ca.mcgill.ecse.dslreasoner.realistic.metrics.calculator.metrics.MetricSampleGroup
+import java.util.HashMap
+import java.util.HashSet
 import java.util.List
+import java.util.Set
 import org.apache.commons.math3.stat.inference.KolmogorovSmirnovTest
 
-class KSDistance {
+class KSDistance extends CostDistance {
         var static ksTester = new KolmogorovSmirnovTest();
-        var double[] mpcSamples;
-        var double[] naSamples;
-        var double[] outDegreeSamples;
-        
+        var MetricSampleGroup g;
         new(Domain d){
                 var metrics = RepMetricsReader.read(d);
-                mpcSamples = metrics.mpcSamples;
-                naSamples = metrics.naSamples.stream.mapToDouble([it]).toArray();
-                outDegreeSamples = metrics.outDegreeSamples.stream.mapToDouble([it]).toArray();
+                this.g = metrics;
         }
         
-        def double mpcDistance(List<Double> samples){
-                // map list to array
-                var arr = samples.stream.mapToDouble([it]).toArray();
-                
+        override double mpcDistance(List<Double> samples){
                 //if the size of array is smaller than 2, ks distance cannot be performed, simply return 1
-                if(arr.size < 2) return 1;
-                return ksTester.kolmogorovSmirnovStatistic(mpcSamples, arr);
+                if(samples.size < 2) return 1;
+                return ksTester.kolmogorovSmirnovStatistic(g.mpcSamples, samples);
         }
         
-        def double naDistance(List<Double> samples){
-                // map list to array
-                var arr = samples.stream.mapToDouble([it]).toArray();
-                
+        override double naDistance(List<Double> samples){
                 //if the size of array is smaller than 2, ks distance cannot be performed, simply return 1
-                if(arr.size < 2) return 1;
-                return ksTester.kolmogorovSmirnovStatistic(naSamples as double[], arr);
+                if(samples.size < 2) return 1;
+                return ksTester.kolmogorovSmirnovStatistic(g.naSamples as double[], samples);
         }
         
-        def double outDegreeDistance(List<Double> samples){
+        override double outDegreeDistance(List<Double> samples){
+                //if the size of array is smaller than 2, ks distance cannot be performed, simply return 1
+                if(samples.size < 2) return 1;
+                return ksTester.kolmogorovSmirnovStatistic(g.outDegreeSamples, samples);
+        }
+        
+        def double typedOutDegreeDistance(HashMap<String, List<Integer>> map){
+                var value = 0.0;
                 // map list to array
-                var arr = samples.stream.mapToDouble([it]).toArray();
+                val keySet = new HashSet<String>(map.keySet);
+                keySet.addAll(g.typedOutDegreeSamples.keySet);
+                for(key : keySet){
+                        if(!map.containsKey(key) ){
+                                value += 1;
+                        }else if(!g.typedOutDegreeSamples.containsKey(key)){
+                                value += map.get(key).size * 100;       
+                        }else{
+                                var double[] rep = g.typedOutDegreeSamples.get(key).stream().mapToDouble([it|it]).toArray();
+                                var double[] ins = map.get(key).stream().mapToDouble([it|it]).toArray();
+                                if((rep.size < 2 || ins.size < 2) ){
+                                        if(rep.size < 2 && rep.containsAll(ins)){
+                                                value += 0;
+                                        }else{
+                                                value += 1;
+                                        }
+                                }else if(rep.size >= 2 && ins.size >= 2){
+                                        value += ksTester.kolmogorovSmirnovStatistic(rep, ins);
+                                }
+                        }
+                } 
                 
-                //if the size of array is smaller than 2, ks distance cannot be performed, simply return 1
-                if(arr.size < 2) return 1;
-                return ksTester.kolmogorovSmirnovStatistic(outDegreeSamples, arr);
+                
+                return value;
         }
 }
\ No newline at end of file