Adding more log messages

src/main/java/com/datumbox/framework/machinelearning/common/bases/featureselection/CategoricalFeatureSelection.java

@@ -28,6 +28,8 @@
 import java.util.Iterator;
 import java.util.List;
 import java.util.Map;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
 
 
@@ -127,7 +129,7 @@ protected void _fit(Dataset data) {
         Map<Object, Double> tmp_featureCounts = dbc.getBigMap("tmp_featureCounts", true); //map which stores the counts of the features
 
 
-        //build the maps with teh feature statistics and counts
+        //build the maps with the feature statistics and counts
         buildFeatureStatistics(data, tmp_classCounts, tmp_featureClassCounts, tmp_featureCounts);
 
 
@@ -151,6 +153,8 @@ protected void filterFeatures(Dataset newdata) {
     }
 
     private static void filterData(Dataset data, DatabaseConnector dbc, Map<Object, Double> featureScores, boolean ignoringNumericalFeatures) {
+        Logger logger = LoggerFactory.getLogger(CategoricalFeatureSelection.class);
+        logger.debug("filterData()");
 
         Map<Object, Boolean> tmp_removedColumns = dbc.getBigMap("tmp_removedColumns", true);
 
@@ -168,7 +172,7 @@ private static void filterData(Dataset data, DatabaseConnector dbc, Map<Object,
             }
         }
 
-
+        logger.debug("Removing Columns");
         data.removeColumns(tmp_removedColumns.keySet());
 
         //Drop the temporary Collection
@@ -189,6 +193,8 @@ public static void removeRareFeatures(Dataset data, DatabaseConnector dbc, Integ
         //feature selection. If called statically, the map should be instatiated
         //just before the call to this method and dropped immediately after 
         //since it has no use.
+        Logger logger = LoggerFactory.getLogger(CategoricalFeatureSelection.class);
+        logger.debug("removeRareFeatures()");
 
         if(!featureCounts.isEmpty()) {
             throw new RuntimeException("The featureCounts map should be empty.");
@@ -197,6 +203,8 @@ public static void removeRareFeatures(Dataset data, DatabaseConnector dbc, Integ
         Map<Object, Dataset.ColumnType> columnTypes = data.getColumns();
 
         //find the featureCounts
+
+        logger.debug("Estimating featureCounts");
         for(Integer rId : data) {
             Record r = data.get(rId);
             for(Map.Entry<Object, Object> entry : r.getX().entrySet()) {
@@ -227,6 +235,7 @@ public static void removeRareFeatures(Dataset data, DatabaseConnector dbc, Integ
 
         //remove rare features
         if(rareFeatureThreshold != null && rareFeatureThreshold>0) {
+            logger.debug("Removing rare features");
             //remove features from the featureCounts list
             Iterator<Map.Entry<Object, Double>> it = featureCounts.entrySet().iterator();
             while(it.hasNext()) {
@@ -236,12 +245,13 @@ public static void removeRareFeatures(Dataset data, DatabaseConnector dbc, Integ
                 }
             }
 
-            //then remove the features in dataset that does not appear in the list
+            //then remove the features in dataset that do not appear in the list
             filterData(data, dbc, featureCounts, ignoringNumericalFeatures);
         }
     }
 
     private void buildFeatureStatistics(Dataset data, Map<Object, Integer> classCounts, Map<List<Object>, Integer> featureClassCounts, Map<Object, Double> featureCounts) {        
+        logger.debug("buildFeatureStatistics()");
         TP trainingParameters = knowledgeBase.getTrainingParameters();
         Integer rareFeatureThreshold = trainingParameters.getRareFeatureThreshold();
         boolean ignoringNumericalFeatures = trainingParameters.isIgnoringNumericalFeatures();
@@ -251,7 +261,9 @@ private void buildFeatureStatistics(Dataset data, Map<Object, Integer> classCoun
         //The map must be empty or else you get a RuntimeException
         removeRareFeatures(data, knowledgeBase.getDbc(), rareFeatureThreshold, featureCounts, ignoringNumericalFeatures);
 
+        Map<Object, Dataset.ColumnType> columnTypes = data.getColumns();
         //now find the classCounts and the featureClassCounts
+        logger.debug("Estimating classCounts and featureClassCounts");
         for(Integer rId : data) {
             Record r = data.get(rId);
             Object theClass = r.getY();
@@ -268,7 +280,7 @@ private void buildFeatureStatistics(Dataset data, Map<Object, Integer> classCoun
                 Object feature = entry.getKey();
 
                 if(ignoringNumericalFeatures) { //if we ignore the numerical features, investigate further if we must skip the feature
-                    if(data.getColumns().get(feature)==Dataset.ColumnType.NUMERICAL) { //is it numerical? 
+                    if(columnTypes.get(feature)==Dataset.ColumnType.NUMERICAL) { //is it numerical? 
                         continue; //skip any further analysis
                     }
                 }

src/main/java/com/datumbox/framework/machinelearning/common/bases/featureselection/ScoreBasedFeatureSelection.java

@@ -20,6 +20,8 @@
 import com.google.common.collect.Ordering;
 import java.util.Iterator;
 import java.util.Map;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
 /**
  * Abstract class which is the base of every Categorical Feature Selection algorithm.
@@ -50,10 +52,14 @@ protected ScoreBasedFeatureSelection(String dbName, DatabaseConfiguration dbConf
 
 
     public static void selectHighScoreFeatures(Map<Object, Double> featureScores, Integer maxFeatures) {
+        Logger logger = LoggerFactory.getLogger(ScoreBasedFeatureSelection.class);
+        logger.debug("selectHighScoreFeatures()");
 
+        logger.debug("Estimating the minPermittedScore");
         Double minPermittedScore=Ordering.<Double>natural().greatestOf(featureScores.values().iterator(), maxFeatures).get(maxFeatures-1);
 
         //remove any entry with score less than the minimum permitted one
+        logger.debug("Removing features with scores less than threshold");
         Iterator<Map.Entry<Object, Double>> it = featureScores.entrySet().iterator();
         while(it.hasNext()) {
             Map.Entry<Object, Double> entry = it.next();
@@ -65,6 +71,7 @@ public static void selectHighScoreFeatures(Map<Object, Double> featureScores, In
         //if some extra features still exist (due to ties on the scores) remove some of those extra features
         int numOfExtraFeatures = featureScores.size()-maxFeatures;
         if(numOfExtraFeatures>0) {
+            logger.debug("Removing extra features caused by ties");
             it = featureScores.entrySet().iterator();
             while(it.hasNext() && numOfExtraFeatures>0) {
                 Map.Entry<Object, Double> entry = it.next();

src/main/java/com/datumbox/framework/machinelearning/featureselection/categorical/ChisquareSelect.java

@@ -63,7 +63,7 @@ public ChisquareSelect(String dbName, DatabaseConfiguration dbConf) {
 
     @Override
     protected void estimateFeatureScores(Map<Object, Integer> classCounts, Map<List<Object>, Integer> featureClassCounts, Map<Object, Double> featureCounts) {
-
+        logger.debug("estimateFeatureScores()");
         ModelParameters modelParameters = knowledgeBase.getModelParameters();
         TrainingParameters trainingParameters = knowledgeBase.getTrainingParameters();
 

src/main/java/com/datumbox/framework/machinelearning/featureselection/categorical/MutualInformation.java

@@ -51,6 +51,7 @@ public MutualInformation(String dbName, DatabaseConfiguration dbConf) {
 
     @Override
     protected void estimateFeatureScores(Map<Object, Integer> classCounts, Map<List<Object>, Integer> featureClassCounts, Map<Object, Double> featureCounts) {
+        logger.debug("estimateFeatureScores()");
         ModelParameters modelParameters = knowledgeBase.getModelParameters();
         TrainingParameters trainingParameters = knowledgeBase.getTrainingParameters();