Conflict rate formula update

datascience · Jan 17, 2024 · eca7671 · eca7671
1 parent 5251238
commit eca7671
Show file tree

Hide file tree

Showing 9 changed files with 47 additions and 26 deletions.
diff --git a/core/src/main/java/rocks/artur/api/GetCollectionStatistics.java b/core/src/main/java/rocks/artur/api/GetCollectionStatistics.java
@@ -5,7 +5,6 @@
 import java.util.Map;
 
 public interface GetCollectionStatistics {
-    Map<String, Double> getSizeStatistics(FilterCriteria filterCriteria);
+    Map<String, Double> getStatistics(FilterCriteria filterCriteria);
 
-    double getConflictRate();
 }
diff --git a/core/src/main/java/rocks/artur/api_impl/GetCollectionStatisticsImpl.java b/core/src/main/java/rocks/artur/api_impl/GetCollectionStatisticsImpl.java
@@ -15,13 +15,9 @@ public GetCollectionStatisticsImpl(CharacterisationResultGateway characterisatio
     }
 
     @Override
-    public Map<String, Double> getSizeStatistics(FilterCriteria filterCriteria) {
-        Map<String, Double> sizeStatistics = characterisationResultGateway.getSizeStatistics(filterCriteria);
+    public Map<String, Double> getStatistics(FilterCriteria filterCriteria) {
+        Map<String, Double> sizeStatistics = characterisationResultGateway.getCollectionStatistics(filterCriteria);
         return sizeStatistics;
     }
 
-    @Override
-    public double getConflictRate() {
-        return characterisationResultGateway.getConflictRate();
-    }
 }
diff --git a/core/src/main/java/rocks/artur/domain/CharacterisationResultGateway.java b/core/src/main/java/rocks/artur/domain/CharacterisationResultGateway.java
@@ -55,7 +55,7 @@ public interface CharacterisationResultGateway {
      */
     List<CharacterisationResult> getConflictsByFilepath(String filepath);
 
-    Map<String, Double> getSizeStatistics(FilterCriteria filterCriteria);
+    Map<String, Double> getCollectionStatistics(FilterCriteria filterCriteria);
 
     List<PropertyValueStatistic> getPropertyValueDistribution(Property property, FilterCriteria<CharacterisationResult> filter);
 

diff --git a/infra-persistence/src/main/java/rocks/artur/jpa/CharacterisationResultGatewayJpaImpl.java b/infra-persistence/src/main/java/rocks/artur/jpa/CharacterisationResultGatewayJpaImpl.java
@@ -158,25 +158,19 @@ public List<CharacterisationResult> getConflictsByFilepath(String filepath) {
     }
 
     @Override
-    public Map<String, Double> getSizeStatistics(FilterCriteria filterCriteria) {
+    public Map<String, Double> getCollectionStatistics(FilterCriteria filterCriteria) {
         Map<String, Double> result = new HashMap<>();
 
         double[] sizeStatistics = characterisationResultViewRepository.getSizeStatistics(filterCriteria);
         result.put("totalSize", sizeStatistics[0]);
-        Long minSize = characterisationResultViewRepository.getMinSize();
         result.put("minSize", sizeStatistics[1]);
-        Long maxSize = characterisationResultViewRepository.getMaxSize();
         result.put("maxSize", sizeStatistics[2]);
-
-        Long avgSize = characterisationResultViewRepository.getAvgSize();
         result.put("avgSize", sizeStatistics[3]);
-
-        Long totalCount = characterisationResultViewRepository.getTotalCount();
         result.put("totalCount", sizeStatistics[4]);
 
-        double conflictRate = this.getConflictRate();
-        result.put("conflictRate", conflictRate);
-
+        double[] conflictStatistics  = characterisationResultViewRepository.getConflictStatistics(filterCriteria);
+        result.put("conflictRate", conflictStatistics[1]);
+        result.put("conflictCount", conflictStatistics[0]);
         return result;
     }
 

diff --git a/...stence/src/main/java/rocks/artur/jpa/view/CustomCharacterisationResultViewRepository.java b/...stence/src/main/java/rocks/artur/jpa/view/CustomCharacterisationResultViewRepository.java
@@ -12,6 +12,8 @@ public interface CustomCharacterisationResultViewRepository {
 
     List<Object[]> getObjects(FilterCriteria filterCriteria);
 
+    double[] getConflictStatistics(FilterCriteria filterCriteria);
+
     List<String[]> getRandomSamples(FilterCriteria filterCriteria, int sampleSize);
 
     List<String[]> getSelectiveFeatureDistributionSamples(FilterCriteria filterCriteria, List<Property> properties);

diff --git a/...ce/src/main/java/rocks/artur/jpa/view/CustomCharacterisationResultViewRepositoryImpl.java b/...ce/src/main/java/rocks/artur/jpa/view/CustomCharacterisationResultViewRepositoryImpl.java
@@ -8,7 +8,6 @@
 import rocks.artur.domain.FilterCriteria;
 import rocks.artur.domain.Property;
 
-import java.util.ArrayList;
 import java.util.List;
 import java.util.stream.Collectors;
 
@@ -82,7 +81,7 @@ public List<Object[]> getObjects(FilterCriteria filterCriteria) {
     }
 
     @Override
-    public double[] getSizeStatistics(FilterCriteria filterCriteria){
+    public double[] getSizeStatistics(FilterCriteria filterCriteria) {
         String subquery = "select distinct FILEPATH from CHARACTERISATIONRESULTVIEW ";
         if (filterCriteria != null) {
             subquery = filterJPA.convert(filterCriteria);
@@ -108,6 +107,38 @@ public double[] getSizeStatistics(FilterCriteria filterCriteria){
         return result;
     }
 
+
+    @Override
+    public double[] getConflictStatistics(FilterCriteria filterCriteria) {
+        String subquery = "select distinct FILEPATH from CHARACTERISATIONRESULTVIEW ";
+        if (filterCriteria != null) {
+            subquery = filterJPA.convert(filterCriteria);
+        }
+
+        String query = String.format(
+                "select count(distinct t.FILEPATH) as count " +
+                        "from CHARACTERISATIONRESULTVIEW t " +
+                        "join (%s) c on t.FILEPATH=c.FILEPATH " +
+                        "where t.PROPERTY_VALUE='CONFLICT'", subquery);
+
+        Long conflictsCount = (Long) entityManager.createNativeQuery(query).getSingleResult();
+
+
+        String query2 = String.format(
+                "select count(distinct t.FILEPATH) as count " +
+                        "from CHARACTERISATIONRESULTVIEW t " +
+                        "join (%s) c on t.FILEPATH=c.FILEPATH ", subquery);
+
+        Long totalCount = (Long) entityManager.createNativeQuery(query2).getSingleResult();
+
+        double rate = 0d;
+        if (totalCount != 0) {
+            rate = (double) conflictsCount / totalCount;
+        }
+        double[] result = new double[]{conflictsCount, rate};
+        return result;
+    }
+
     @Override
     public List<String[]> getRandomSamples(FilterCriteria filterCriteria, int sampleSize) {
         String subquery = "select distinct FILEPATH from CHARACTERISATIONRESULTVIEW ";
@@ -181,5 +212,4 @@ public List<String[]> getSelectiveFeatureDistributionSamples(FilterCriteria filt
     }
 
 
-
 }
diff --git a/infra-rest/src/main/java/rocks/artur/endpoints/RestService.java b/infra-rest/src/main/java/rocks/artur/endpoints/RestService.java
@@ -104,7 +104,7 @@ public List<Property> getConflictsPerObject(
     public Map<String, Double> getCollectionStatistics(@RequestParam(name = "filter", required = false) @Parameter(name = "filter", description = "Filter", example = "FORMAT=\"Portable Document Format\"") String filter) throws ParseException {
         CriteriaParser parser = new CriteriaParser();
         FilterCriteria filterCriteria = parser.parse(filter);
-        Map<String, Double> sizeStatistics = getCollectionStatistics.getSizeStatistics(filterCriteria);
+        Map<String, Double> sizeStatistics = getCollectionStatistics.getStatistics(filterCriteria);
         return sizeStatistics;
     }
 

diff --git a/main/src/test/java/rocks/artur/CharacterisationResultGatewayImplTest.java b/main/src/test/java/rocks/artur/CharacterisationResultGatewayImplTest.java
@@ -102,7 +102,7 @@ void getCollectionStatisticsWithoutFilterTest() throws ParseException {
         String typeFilter = "FORMAT=\"Portable Document Format\"";
         CriteriaParser parser = new CriteriaParser();
         FilterCriteria parse = parser.parse(typeFilter);
-        Map<String, Double> sizeStatistics = characterisationResultGatewaySqlImpl.getSizeStatistics(null);
+        Map<String, Double> sizeStatistics = characterisationResultGatewaySqlImpl.getCollectionStatistics(null);
         Assert.assertEquals(10047.0, sizeStatistics.get("totalSize"), 0.1);
         System.out.println(sizeStatistics);
     }
@@ -112,7 +112,7 @@ void getCollectionStatisticsWithFilterTest() throws ParseException {
         String typeFilter = "FORMAT=\"Portable Document Format\"";
         CriteriaParser parser = new CriteriaParser();
         FilterCriteria parse = parser.parse(typeFilter);
-        Map<String, Double> sizeStatistics = characterisationResultGatewaySqlImpl.getSizeStatistics(parse);
+        Map<String, Double> sizeStatistics = characterisationResultGatewaySqlImpl.getCollectionStatistics(parse);
         Assert.assertEquals(4.0, sizeStatistics.get("totalSize"), 0.1);
         System.out.println(sizeStatistics);
     }

diff --git a/web/frontend/src/scenes/PropertyValueDistribution/index.jsx b/web/frontend/src/scenes/PropertyValueDistribution/index.jsx
@@ -63,7 +63,7 @@ const PropertyValueDistribution = (payload) => {
   }, [filter]);
 
   let filterClick = (property, event) => {
-    if (event.indexValue == "...others") {
+    if (event.indexValue == ".etc") {
       return;
     }
     let newCondition = null;