3 files changed, 42 insertions, 20 deletions

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
index b945d97b..d6adcc9a 100644
--- 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
@@ -18,13 +18,12 @@ class EuclideanDistance extends CostDistance{
         var HashMap<String, Double> outDegreePMF;
         var DecimalFormat formatter;
         
-        new(Domain d){
-                var metrics = RepMetricsReader.read(d);
-                this.g = metrics;
+        new(MetricSampleGroup g){
+                this.g = g;
                 
-                var mpcSamples = metrics.mpcSamples;
-                var naSamples = metrics.naSamples.stream.mapToDouble([it]).toArray();
-                var outDegreeSamples = metrics.outDegreeSamples.stream.mapToDouble([it]).toArray();
+                var mpcSamples = g.mpcSamples;
+                var naSamples = g.naSamples.stream.mapToDouble([it]).toArray();
+                var outDegreeSamples = g.outDegreeSamples.stream.mapToDouble([it]).toArray();
                 
                 //needs to format the number to string avoid precision issue
                 formatter = new DecimalFormat("#0.00000");  
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 df65b81f..4a0a0dc3 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
@@ -1,7 +1,6 @@
 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.HashMap
@@ -11,13 +10,13 @@ class JSDistance extends CostDistance {
         var HashMap<String, Double> mpcPMF;
         var HashMap<String, Double> naPMF;
         var HashMap<String, Double> outDegreePMF;
+        var HashMap<String, Double> nodeTypesPMF;
         var DecimalFormat formatter;
         
-        new(Domain d){
-                var metrics = RepMetricsReader.read(d);
-                var mpcSamples = metrics.mpcSamples;
-                var naSamples = metrics.naSamples.stream.mapToDouble([it]).toArray();
-                var outDegreeSamples = metrics.outDegreeSamples.stream.mapToDouble([it]).toArray();
+        new(MetricSampleGroup g){
+                var mpcSamples = g.mpcSamples;
+                var naSamples = g.naSamples.stream.mapToDouble([it]).toArray();
+                var outDegreeSamples = g.outDegreeSamples.stream.mapToDouble([it]).toArray();
                 
                 //needs to format the number to string avoid precision issue
                 formatter = new DecimalFormat("#0.00000");  
@@ -25,6 +24,7 @@ class JSDistance extends CostDistance {
                 mpcPMF = pmfFromSamples(mpcSamples, formatter);
                 naPMF = pmfFromSamples(naSamples, formatter);
                 outDegreePMF = pmfFromSamples(outDegreeSamples, formatter);     
+                nodeTypesPMF = g.nodeTypeSamples; 
         }
         
         def private combinePMF(HashMap<String, Double> pmf1, HashMap<String, Double> pmf2){
@@ -81,4 +81,8 @@ class JSDistance extends CostDistance {
                 if(map.size < 2) return 1;
                 return jsDivergence(map, outDegreePMF);
         }
+        
+        def nodeTypeDistance(HashMap<String, Double> samples){
+                return klDivergence(samples, nodeTypesPMF);
+        }
 }
\ 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/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 86f5f23c..08d8704a 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
@@ -1,22 +1,18 @@
 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 extends CostDistance {
         var static ksTester = new KolmogorovSmirnovTest();
         var MetricSampleGroup g;
-        new(Domain d){
-                var metrics = RepMetricsReader.read(d);
-                this.g = metrics;
-        }
         
+        new(MetricSampleGroup g){
+                this.g = g;     
+        }
         override double mpcDistance(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;
@@ -63,4 +59,27 @@ class KSDistance extends CostDistance {
                 
                 return value;
         }
+        
+        def nodeTypeDistance(HashMap<String, Double> samples){
+                var typesDistMap = g.nodeTypeSamples;
+                var sourceDist = newArrayList();
+                var instanceDist = newArrayList();
+                
+                for(key : typesDistMap.keySet()){
+                        sourceDist.add(typesDistMap.get(key));
+                        instanceDist.add(samples.getOrDefault(key, 0.0));
+                }
+                
+                // Since we already know the pdf, we compute the ks-test manully
+                var ksStatistics = 0.0;
+                var sum1 = 0.0;
+                var sum2 = 0.0;
+                for(var i = 0; i < sourceDist.size(); i++){
+                        sum1 += sourceDist.get(i);
+                        sum2 += instanceDist.get(i);
+                        
+                        ksStatistics = Math.max(ksStatistics, Math.abs(sum1 - sum2));
+                } 
+                return ksStatistics;
+        }
 }
\ No newline at end of file