Automated initialization of tabular data sets #13

AutoTuning/src/main/java/RandomSearch/ContinuousParameter.java

@@ -13,12 +13,22 @@ public class ContinuousParameter implements ContinuousParameterInterface {
     private double currentValue;
 
     public ContinuousParameter(String name, double defaultValue, double minValue, double maxValue) {
+        this(name, defaultValue, minValue, maxValue, 0);
+    }
+
+    public ContinuousParameter(String name, double currentValue) {
+        this(name, -1, -1, -1 , currentValue);
+    }
+
+    public ContinuousParameter(String name, double defaultValue, double minValue, double maxValue, double currentValue) {
         this.name = name;
         this.defaultValue = defaultValue;
         this.minValue = minValue;
         this.maxValue = maxValue;
+        this.currentValue = currentValue;
     }
 
+
     public String getType() {
         return type;
     }

AutoTuning/src/main/java/RandomSearch/HyperparameterSpace.java

@@ -1,7 +1,6 @@
 package RandomSearch;
 
 import de.viadee.xai.anchor.adapter.tabular.TabularInstance;
-import scala.reflect.internal.Trees;
 
 import java.util.ArrayList;
 import java.util.List;
@@ -14,16 +13,34 @@ public class HyperparameterSpace {
     private double coverage = 0;
     private long runtime = 0;
 
-    private List<Parameter> hyperParameters = new ArrayList<Parameter>();
+    private List<Parameter> hyperParameters;
 
 
     public HyperparameterSpace() {
-        this.hyperParameters.add(new IntegerParameter("beamsize", 2, 1, 30));
-        this.hyperParameters.add(new ContinuousParameter("tau", 1, 0.1, 1.0));
-        this.hyperParameters.add(new ContinuousParameter("delta", 0.1, 0.1, 0.5));
-        this.hyperParameters.add(new ContinuousParameter("epsilon", 0.1, 0.1, 0.5));
-        this.hyperParameters.add(new ContinuousParameter("tauDiscrepancy", 0.05, 0.01, 0.1));
-        this.hyperParameters.add(new IntegerParameter("initSampleCount", 1, 1, 10));
+        this(null);
+    }
+
+    public HyperparameterSpace(List<Parameter> hyperParameters) {
+
+        if (hyperParameters != null) {
+            this.hyperParameters = hyperParameters;
+        } else {
+            this.hyperParameters = fillWithDefaults();
+        }
+    }
+
+    private List<Parameter> fillWithDefaults() {
+
+        List<Parameter> parameters = new ArrayList<Parameter>();
+
+        parameters.add(new IntegerParameter("beamsize", 2, 1, 30));
+        parameters.add(new ContinuousParameter("tau", 1, 0.1, 1.0));
+        parameters.add(new ContinuousParameter("delta", 0.1, 0.1, 0.5));
+        parameters.add(new ContinuousParameter("epsilon", 0.1, 0.1, 0.5));
+        parameters.add(new ContinuousParameter("tauDiscrepancy", 0.05, 0.01, 0.1));
+        parameters.add(new IntegerParameter("initSampleCount", 1, 1, 10));
+
+        return parameters;
     }
 
     public void setRandomHyperparameterSpace() {

AutoTuning/src/main/java/RandomSearch/IntegerParameter.java

@@ -10,14 +10,26 @@ public class IntegerParameter implements IntegerParameterInterface {
     private int maxValue;
     private int currentValue;
 
+
     public IntegerParameter(String name, int defaultValue, int minValue, int maxValue) {
+        this(name, defaultValue, minValue, maxValue, -1);
+    }
+
+    public IntegerParameter(String name, int currentValue) {
+        this(name, -1, -1, -1, currentValue);
+    }
+
+    public IntegerParameter(String name, int defaultValue, int minValue, int maxValue, int currentValue) {
         this.name = name;
         this.defaultValue = defaultValue;
         this.minValue = minValue;
         this.maxValue = maxValue;
+        this.currentValue = currentValue;
     }
 
-    public String getType() { return type; }
+    public String getType() {
+        return type;
+    }
 
     public String getName() {
         return name;

AutoTuning/src/main/java/TitanicDataset.java

@@ -1,3 +1,5 @@
+package data;
+
 import de.viadee.xai.anchor.adapter.tabular.AnchorTabular;
 import de.viadee.xai.anchor.adapter.tabular.column.DoubleColumn;
 import de.viadee.xai.anchor.adapter.tabular.column.IntegerColumn;
@@ -23,7 +25,7 @@ public class TitanicDataset {
     /**
      * @return the {@link AnchorTabular} object that contains the training data and its definitions
      */
-    static AnchorTabular createTabularTrainingDefinition() {
+    public static AnchorTabular createTabularTrainingDefinition() {
         InputStream trainingDataStream = ClassLoader.getSystemResourceAsStream("Titanic/train.csv");
         if (trainingDataStream == null)
             throw new RuntimeException("Could not load data");
@@ -56,7 +58,7 @@ static AnchorTabular createTabularTrainingDefinition() {
     /**
      * @return the {@link AnchorTabular} object that contains the test data and its definitions
      */
-    static AnchorTabular createTabularTestDefinition() {
+    public static AnchorTabular createTabularTestDefinition() {
         // The following implementation is very much similar to the above method.
         // It is contained in an own block to increase the tutorial's readability
         // Main difference: no target label is included in test set data
@@ -103,7 +105,7 @@ public Serializable apply(Serializable serializable) {
     /**
      * @return the labels of the test set as specified in gender_submission.
      */
-    static int[] readTestLabels() {
+    public static int[] readTestLabels() {
         InputStream trainingDataStream = ClassLoader.getSystemResourceAsStream("Titanic/gender_submission.csv");
         if (trainingDataStream == null)
             throw new RuntimeException("Could not load data");

AutoTuning/src/test/java/Initializer/FileInitializerTest.java

@@ -0,0 +1,49 @@
+package Initializer;
+
+import org.junit.Assert;
+import org.junit.Test;
+
+public class FileInitializerTest {
+
+    @Test(expected = RuntimeException.class)
+    public void testEmptyPath() {
+
+        // Given
+        String path = "";
+
+        // When
+        FileInitializer fi = new FileInitializer(path);
+    }
+
+    @Test
+    public void testNotExistingFile() {
+
+        // Given
+        String path = "test";
+
+        // When
+        FileInitializer fi = new FileInitializer(path);
+        fi.setExtension();
+        String extension = fi.getExtension();
+
+        // Then
+        Assert.assertFalse(extension == null);
+    }
+
+    @Test
+    public void testExcelFile() {
+
+        // Given
+        String path = "src/test/resources/train.csv";
+
+        // When
+        FileInitializer fi = new FileInitializer(path);
+        fi.setExtension();
+        String extension = fi.getExtension();
+
+        // Then
+        Assert.assertTrue(extension.equals("csv"));
+
+    }
+
+}

AutoTuning/src/test/java/Initializer/TabularInitializerTest.java

@@ -0,0 +1,87 @@
+package Initializer;
+
+import de.viadee.xai.anchor.adapter.tabular.util.CSVReader;
+import org.junit.Assert;
+import org.junit.Test;
+
+import java.io.IOException;
+import java.io.InputStream;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Collection;
+
+public class TabularInitializerTest {
+
+    @Test
+    public void createTabularTrainingDefinition() {
+
+        // Given
+        String path = "train.csv";
+
+        // Then
+        TabularInitializer.createTabularDefinition(path, 0, true);
+
+    }
+
+    @Test
+    public void testGetColumnDatatype() {
+
+        // Given
+        String path = "train.csv";
+        InputStream trainingDataStream = ClassLoader.getSystemResourceAsStream(path);
+        Collection<String[]> strings = null;
+        try {
+            strings = CSVReader.readCSV(trainingDataStream, false);
+        } catch (IOException e) {
+
+        }
+
+        // When
+        int[] result = TabularInitializer.getColumnDataTypes(strings);
+
+    }
+
+    @Test
+    public void testEmptyRowDatatypes() {
+        // Given
+        String[] row = {};
+        Collection<String[]> dataframe = new ArrayList<>();
+        dataframe.add(row);
+
+        // When
+        int[] result = TabularInitializer.getColumnDataTypes(dataframe);
+
+        // Then
+        Assert.assertTrue(result.length == 0);
+    }
+
+    @Test
+    public void testIntRowDatatypes() {
+        // Given
+        String[] row = {"1", "2", "3", "4"};
+        Collection<String[]> dataframe = new ArrayList<>();
+        dataframe.add(row);
+
+        // When
+        int[] result = TabularInitializer.getColumnDataTypes(dataframe);
+        int[] test ={0, 0, 0, 0};
+
+        // Then
+        Assert.assertTrue(Arrays.equals(result,test));
+    }
+
+    @Test
+    public void testAllDatatypes() {
+        // Given
+        String[] row = new String[]{"1", "Hallo", "Test123.3", "3.12"};
+        Collection<String[]> dataframe = new ArrayList<>();
+        dataframe.add(row);
+
+        // When
+        int[] result = TabularInitializer.getColumnDataTypes(dataframe);
+        int[] test ={0, 2, 2, 1};
+
+        // Then
+        Assert.assertTrue(Arrays.equals(result,test));
+    }
+}