Replaced Map<Object, Object> with Map<Object, Double> when possible.

datumbox · Dec 28, 2016 · 81ebda1 · 81ebda1
1 parent e5e6377
commit 81ebda1
Show file tree

Hide file tree

Showing 4 changed files with 8 additions and 9 deletions.
diff --git a/...ork-core/src/main/java/com/datumbox/framework/core/machinelearning/clustering/Kmeans.java b/...ork-core/src/main/java/com/datumbox/framework/core/machinelearning/clustering/Kmeans.java
@@ -712,8 +712,8 @@ else if(initializationMethod==TrainingParameters.Initialization.PLUS_PLUS) {
             StorageEngine storageEngine = knowledgeBase.getStorageEngine();
             Set<Integer> alreadyAddedPoints = new HashSet(); //this is small. equal to k
             for(int i = 0; i < k; ++i) {
-                Map<Object, Object> tmp_minClusterDistance = storageEngine.getBigMap("tmp_minClusterDistance", Object.class, Object.class, MapType.HASHMAP, StorageHint.IN_MEMORY, true, true);
-                AssociativeArray minClusterDistanceArray = new AssociativeArray(tmp_minClusterDistance);
+                Map<Object, Double> tmp_minClusterDistance = storageEngine.getBigMap("tmp_minClusterDistance", Object.class, Double.class, MapType.HASHMAP, StorageHint.IN_MEMORY, true, true);
+                AssociativeArray minClusterDistanceArray = new AssociativeArray((Map)tmp_minClusterDistance);
 
                 streamExecutor.forEach(StreamMethods.stream(trainingData.entries(), isParallelized()), e -> {
                     Integer rId = e.getKey();
@@ -739,7 +739,6 @@ else if(initializationMethod==TrainingParameters.Initialization.PLUS_PLUS) {
                 Integer selectedRecordId = (Integer) SimpleRandomSampling.weightedSampling(minClusterDistanceArray, 1, true).iterator().next();
 
                 storageEngine.dropBigMap("tmp_minClusterDistance", tmp_minClusterDistance);
-                //minClusterDistanceArray = null;
 
 
                 alreadyAddedPoints.add(selectedRecordId);

diff --git a/...com/datumbox/framework/core/machinelearning/common/abstracts/algorithms/AbstractDPMM.java b/...com/datumbox/framework/core/machinelearning/common/abstracts/algorithms/AbstractDPMM.java
@@ -31,10 +31,10 @@
 import com.datumbox.framework.core.statistics.descriptivestatistics.Descriptives;
 import com.datumbox.framework.core.statistics.sampling.SimpleRandomSampling;
 
+import java.util.HashMap;
 import java.util.Map;
 import java.util.Objects;
 import java.util.Set;
-import java.util.concurrent.ConcurrentHashMap;
 
 
 /**
@@ -527,7 +527,7 @@ private int collapsedGibbsSampling(Dataframe dataset) {
     }
 
     private AssociativeArray clusterProbabilities(Record r, int n, Map<Integer, CL> clusterMap) {
-        Map<Object, Object> condProbCiGivenXiAndOtherCi = new ConcurrentHashMap<>();
+        Map<Integer, Double> condProbCiGivenXiAndOtherCi = new HashMap<>();
         double alpha = knowledgeBase.getTrainingParameters().getAlpha();
 
         //Probabilities that appear on https://www.cs.cmu.edu/~kbe/dp_tutorial.pdf
@@ -543,7 +543,7 @@ private AssociativeArray clusterProbabilities(Record r, int n, Map<Integer, CL>
             condProbCiGivenXiAndOtherCi.put(clusterId, marginalLogLikelihoodXi+Math.log(mixingXi)); //concurrent map and non-overlapping keys for each thread
         }
 
-        return new AssociativeArray(condProbCiGivenXiAndOtherCi);
+        return new AssociativeArray((Map)condProbCiGivenXiAndOtherCi);
     }
 
     private Object getSelectedClusterFromScores(AssociativeArray clusterScores) {

diff --git a/...a/com/datumbox/framework/core/machinelearning/preprocessing/CornerConstraintsEncoder.java b/...a/com/datumbox/framework/core/machinelearning/preprocessing/CornerConstraintsEncoder.java
@@ -109,7 +109,7 @@ protected void _fit(Dataframe trainingData) {
             for(Map.Entry<Object, Object> entry: r.getX().entrySet()) {
                 Object column = entry.getKey();
                 if(covert2dummy(columnTypes.get(column))) {
-                    referenceLevels.putIfAbsent(column, entry.getValue()); //This Map is an implementation of ConcurrentHashMap and we don't need a synchronized is needed.
+                    referenceLevels.putIfAbsent(column, entry.getValue()); //This Map is thread safe and we don't need a synchronized is needed.
                 }
             }
         }

diff --git a/...src/test/java/com/datumbox/framework/core/statistics/descriptivestatistics/RanksTest.java b/...src/test/java/com/datumbox/framework/core/statistics/descriptivestatistics/RanksTest.java
@@ -21,7 +21,7 @@
 import org.junit.Test;
 
 import java.util.Arrays;
-import java.util.concurrent.ConcurrentSkipListMap;
+import java.util.HashMap;
 
 import static org.junit.Assert.assertEquals;
 
@@ -40,7 +40,7 @@ public void testGetRanksFromValues() {
         logger.info("getRanksFromValues");
         FlatDataList flatDataCollection =  new FlatDataList(Arrays.asList(new Object[]{50,10,10,30,40}));
         FlatDataList expResult = new FlatDataList(Arrays.asList(new Object[]{5.0,1.5,1.5,3.0,4.0}));
-        AssociativeArray expResult2 = new AssociativeArray(new ConcurrentSkipListMap<>());
+        AssociativeArray expResult2 = new AssociativeArray(new HashMap<>());
         expResult2.put(10, 2);
         AssociativeArray tiesCounter = Ranks.getRanksFromValues(flatDataCollection);
         assertEquals(expResult, flatDataCollection);