Removed strange mincov adjustment (results may change).

Author: agudys

adaa.analytics.rules/build.gradle

@@ -27,7 +27,7 @@ codeQuality {
 }
 
 sourceCompatibility = 1.8
-version = '1.7.10'
+version = '1.7.11'
 
 
 jar {

adaa.analytics.rules/src/main/java/adaa/analytics/rules/logic/induction/AbstractFinder.java

@@ -358,10 +358,7 @@ protected Set<Attribute> names2attributes(Set<String> names, ExampleSet dataset)
 			return fractionCurrentGeneration * sizeToCover;
 		}else
 		{
-			return Math.min(
-					params.getAbsoluteMinimumCovered(size),
-					Math.max(1.0, 0.2 * size));
-
+			return params.getAbsoluteMinimumCovered(size);
 		}
 	}
 

adaa.analytics.rules/src/main/java/adaa/analytics/rules/logic/induction/ActionFinder.java

@@ -232,9 +232,7 @@ public boolean tryAddCondition(
 			}
 
 			// analyse stopping criteria
-			double adjustedMinCov = Math.min(
-					params.getMinimumCovered(),
-					Math.max(1.0, 0.2 * ct.weighted_P));
+			double adjustedMinCov = params.getMinimumCovered();
 
 			if (ct.weighted_p < adjustedMinCov) {
 				if (rule.getPremise().getSubconditions().size() == 0) {

adaa.analytics.rules/src/main/java/adaa/analytics/rules/logic/induction/ApproximateClassificationSnC.java

@@ -19,7 +19,6 @@
 public class ApproximateClassificationSnC extends ClassificationSnC {
     public ApproximateClassificationSnC(AbstractFinder finder, InductionParameters params) {
         super(finder, params);
-        numClassThreads = 1; // fix number of class threads
     }
 
     @Override

adaa.analytics.rules/src/main/java/adaa/analytics/rules/logic/induction/ClassificationExpertFinder.java

@@ -361,16 +361,10 @@ public int grow(
 	 */
 	@Override
 	protected boolean checkCandidate(ElementaryCondition cnd, double classId, double p, double n, double new_p, double P,double uncoveredSize,  int ruleOrderNum) {
-		double adjustedMinCov =
-				Math.min(
-						params.getAbsoluteMinimumCovered(P),
-						Math.max(1.0, 0.2 * P));
-		if (new_p >= adjustedMinCov && p >= params.getAbsoluteMinimumCoveredAll(P)) {
-			return true &&
-					!knowledge.isForbidden(cnd.getAttribute(), cnd.getValueSet(), (int)classId);
-		} else {
-			return false;
-		}
+
+		return new_p >= params.getAbsoluteMinimumCovered(P)
+				&& p >= params.getAbsoluteMinimumCoveredAll(P)
+				&& !knowledge.isForbidden(cnd.getAttribute(), cnd.getValueSet(), (int)classId);
 	}
 
 

adaa.analytics.rules/src/main/java/adaa/analytics/rules/logic/induction/ClassificationFinder.java

@@ -906,13 +906,9 @@ public boolean tryAddCondition(
 	 * @return
 	 */
 	protected boolean checkCandidate(ElementaryCondition cnd, double classId, double p, double n, double new_p, double P,double uncoveredSize, int ruleOrderNum) {
-		double adjustedMinCov =
-				countAbsoluteMinimumCovered(P, ruleOrderNum, uncoveredSize);
-		if (new_p >= adjustedMinCov && p >= params.getAbsoluteMinimumCoveredAll(P)) {
-			return true;
-		} else {
-			return false;
-		}
+
+		return new_p >= countAbsoluteMinimumCovered(P, ruleOrderNum, uncoveredSize)
+				&& p >= params.getAbsoluteMinimumCoveredAll(P);
 	}
 
 

adaa.analytics.rules/src/main/java/adaa/analytics/rules/logic/induction/ClassificationSnC.java

@@ -45,8 +45,6 @@ public class ClassificationSnC extends AbstractSeparateAndConquer {
 	 */
 	protected AbstractFinder finder;
 
-	protected int numClassThreads = 1;
-
 	public ClassificationSnC(AbstractFinder finder, InductionParameters params) {
 		super(params);
 		this.finder = finder;

adaa.analytics.rules/src/main/java/adaa/analytics/rules/logic/induction/InductionParameters.java

@@ -77,7 +77,7 @@ public class InductionParameters implements Serializable {
 
 	public double getMinimumCovered() {return minimumCovered;}
 	public double getAbsoluteMinimumCovered(double size) {
-		return minimumCovered * (minimumCovered >= 1 ? 1 : size);
+		return Math.max(1, minimumCovered * (minimumCovered >= 1 ? 1 : size));
 	}
 	public void setMinimumCovered(double minimumCovered) {this.minimumCovered = minimumCovered;}
 

adaa.analytics.rules/test/resources/reports/ClassificationExpertSnCTest/test_seismic-bumps.guided-c2, gimpuls_750 extended.seismic-bumps-train-minimal.txt

@@ -3,6 +3,4 @@
 Rules
 	IF [gimpuls = (-inf, 1350.50)] THEN class = {0}
 	IF goenergy = <-32.50, inf) AND gimpuls = (-inf, 2784) AND goimpuls = <-39, 111) AND nbumps = (-inf, 5.50) THEN class = {0}
-	IF [gimpuls = <1350.50, inf)] AND genergy = (-inf, 344875) AND senergy = (-inf, 4650) THEN class = {1}
-	IF ghazard = {a} AND gimpuls = (-inf, 1758.50) AND goimpuls = (-inf, 10.50) AND genergy = <17495, inf) AND senergy = <2650, 8250) AND nbumps2 = <0.50, inf) THEN class = {1}
-	IF ghazard = {a} AND goenergy = <-21, 9) AND gimpuls = (-inf, 688.50) AND goimpuls = (-inf, 17.50) AND genergy = <24830, 36505) THEN class = {1}
+	IF [gimpuls = <408.50, inf)] AND ghazard = {a} AND goenergy = (-inf, 94.50) AND maxenergy = (-inf, 6500) AND genergy = <35165, 294530) AND senergy = (-inf, 8750) AND nbumps3 = (-inf, 2.50) THEN class = {1}

adaa.analytics.rules/test/resources/reports/ClassificationExpertSnCTest/test_seismic-bumps.guided-c3, gimpuls_genergy_senergy.seismic-bumps-train-minimal.txt

@@ -5,6 +5,4 @@ Rules
 	IF [gimpuls = (-inf, 1350.50)] AND [senergy = (-inf, 5750)] AND goenergy = <-73.50, inf) AND nbumps = (-inf, 2.50) THEN class = {0}
 	IF [gimpuls = (-inf, 1350.50)] AND [genergy = (-inf, 122140)] AND goenergy = <-73.50, inf) AND nbumps = (-inf, 5.50) THEN class = {0}
 	IF [gimpuls = (-inf, 2784)] AND [senergy = (-inf, 5750)] AND goenergy = <-73.50, inf) AND goimpuls = (-inf, 107.50) THEN class = {0}
-	IF [gimpuls = <1350.50, inf)] AND [genergy = (-inf, 344875)] AND senergy = (-inf, 4650) THEN class = {1}
-	IF [genergy = <18890, inf)] AND [senergy = <950, inf)] AND ghazard = {a} AND goenergy = (-inf, 52.50) AND maxenergy = <1450, 45000) AND nbumps3 = (-inf, 2.50) AND nbumps2 = <0.50, inf) THEN class = {1}
-	IF [genergy = <18890, inf)] AND [gimpuls = <174.50, inf)] AND ghazard = {a} AND goenergy = (-inf, 9) AND maxenergy = (-inf, 6500) AND goimpuls = <-72, 17.50) THEN class = {1}
+	IF [genergy = <18890, inf)] AND [senergy = <950, inf)] AND ghazard = {a} AND goenergy = (-inf, 12) AND maxenergy = (-inf, 6500) AND gimpuls = (-inf, 1845.50) AND goimpuls = (-inf, 10.50) AND shift = {W} AND nbumps3 = <0.50, inf) THEN class = {1}

adaa.analytics.rules/test/resources/reports/ClassificationExpertSnCTest/test_seismic-bumps.guided-c4, seismic OR seismoacoustic OR ghazard.seismic-bumps-train-minimal.txt

@@ -9,7 +9,4 @@ Rules
 	IF [seismoacoustic = {a}] AND gimpuls = (-inf, 1350.50) AND goimpuls = <-39.50, inf) THEN class = {0}
 	IF genergy = (-inf, 25125) AND nbumps = (-inf, 1.50) THEN class = {0}
 	IF goenergy = <-73.50, inf) AND gimpuls = (-inf, 1350.50) AND nbumps = (-inf, 2.50) THEN class = {0}
-	IF [seismic = {b}] AND gimpuls = (-inf, 1878) AND genergy = <117615, 225545) THEN class = {1}
-	IF goenergy = <-41, inf) AND gimpuls = <1350.50, inf) AND maxenergy = (-inf, 45000) AND genergy = (-inf, 1158410) AND nbumps = (-inf, 3.50) THEN class = {1}
-	IF ghazard = {a} AND gimpuls = (-inf, 1758.50) AND goimpuls = (-inf, 10.50) AND genergy = <17495, inf) AND senergy = <2650, 8250) AND nbumps2 = <0.50, inf) THEN class = {1}
-	IF ghazard = {a} AND goenergy = <-21, 9) AND gimpuls = (-inf, 688.50) AND goimpuls = (-inf, 17.50) AND genergy = <24830, 36505) THEN class = {1}
+	IF [seismic = {b}] AND ghazard = {a} AND goenergy = <-39, 94.50) AND maxenergy = (-inf, 45000) AND genergy = <25125, inf) AND goimpuls = <-15.50, inf) THEN class = {1}

adaa.analytics.rules/test/resources/reports/ClassificationExpertSnCTest/test_seismic-bumps.guided-c5, +seismic +nbumps +energy +maxenergy +shift.seismic-bumps-train-minimal.txt

@@ -5,7 +5,4 @@ Rules
 	IF goenergy = <-73.50, inf) AND nbumps = (-inf, 1.50) THEN class = {0}
 	IF goenergy = <-73.50, inf) AND genergy = (-inf, 117615) AND nbumps2 = (-inf, 1.50) THEN class = {0}
 	IF nbumps = (-inf, 5.50) THEN class = {0}
-	IF senergy = <2650, 3850) AND nbumps = <1.50, inf) THEN class = {1}
-	IF goenergy = (-inf, 11) AND genergy = <17495, 401025) AND senergy = <2650, 8250) AND nbumps2 = <0.50, inf) THEN class = {1}
-	IF goenergy = <-21, 9) AND genergy = <24830, 36280) AND senergy = (-inf, 1500) AND nbumps2 = (-inf, 0.50) THEN class = {1}
-	IF goenergy = (-inf, 50) AND maxenergy = (-inf, 45000) AND genergy = <62050, 1158410) AND nbumps = (-inf, 3.50) THEN class = {1}
+	IF goenergy = (-inf, 9) AND maxenergy = (-inf, 6500) AND genergy = <18890, 314675) AND shift = {W} AND nbumps3 = <0.50, inf) THEN class = {1}

adaa.analytics.rules/test/resources/reports/ClassificationExpertSnCTest/test_seismic-bumps.guided-c6, +seismoacoustic +shift +genergy +gimpuls +genergy +goimpuls +ghazard.seismic-bumps-train-minimal.txt

@@ -3,7 +3,4 @@
 Rules
 	IF gimpuls = (-inf, 1350.50) THEN class = {0}
 	IF goenergy = <-32.50, inf) AND gimpuls = (-inf, 2784) AND goimpuls = <-39, 111) THEN class = {0}
-	IF gimpuls = <1350.50, 1507.50) AND genergy = (-inf, 792375) THEN class = {1}
-	IF genergy = <18890, 27830) AND goimpuls = (-inf, -35.50) THEN class = {1}
-	IF goenergy = <5, 9) AND genergy = <18890, inf) AND goimpuls = <1.50, inf) THEN class = {1}
-	IF goenergy = (-inf, -15) AND gimpuls = <387, inf) AND genergy = <24830, inf) AND goimpuls = (-inf, -5.50) AND seismoacoustic = {a} THEN class = {1}
+	IF ghazard = {a} AND goenergy = <-38.50, 9) AND gimpuls = <263, 1878) AND genergy = <24830, 314675) AND goimpuls = (-inf, 17.50) AND shift = {W} THEN class = {1}