[BUG] Taking into account if criterion is benefit (#7)

ldss-hse · May 6, 2022 · bc558e7 · bc558e7
1 parent 858dd67
commit bc558e7
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 17 deletions.
diff --git a/src/main/java/dss/lingvo/samples/TT2HFLTSCoordinator.java b/src/main/java/dss/lingvo/samples/TT2HFLTSCoordinator.java
@@ -67,16 +67,14 @@ private void processSimpleCase(File outputDirectory) throws IOException {
 
     private void processNumericSample() {
         // now how to work with numbers
-        int targetScaleSize = 7;
-
         final float numeric_assessment = 0.78f;
         List<Float> fSet = TTNormalizedTranslator.getInstance().getFuzzySetForNumericEstimation(numeric_assessment, 5);
         float resTranslation = TTNormalizedTranslator.getInstance().getTranslationFromFuzzySet(fSet);
         TTTuple resTuple = TTNormalizedTranslator.getInstance().getTTupleForNumericTranslation(resTranslation, 5);
         System.out.printf("Translating %f to 2-tuple: %s\n", numeric_assessment, resTuple.toString());
     }
 
-    private void processweightsGenerationSample() {
+    private void processWeightsGenerationSample() {
         float[] expDistr = {0.8f, 0.2f};
         float[] resVector = TTUtils.calculateWeightsVector(expDistr, 4);
         System.out.println(Arrays.toString(resVector));
@@ -111,13 +109,22 @@ private void processMultiLevelAdvancedSample(File inputFile, File outputDirector
         int targetScaleSize = 7;
 
         TTJSONMultiLevelInputModel model = ttjsonReader.readJSONMultiLevelDescription(inputFile, false);
-        List<ArrayList<ArrayList<TT2HFLTS>>> all = TTUtils.getAllEstimationsFromMultiLevelJSONModel(model, 7);
+
+        TTNormalizedTranslator.registerScalesBatch(model.getScales());
+
+        List<ArrayList<ArrayList<TT2HFLTS>>> all = TTUtils.getAllEstimationsFromMultiLevelJSONModel(model,
+                targetScaleSize);
 
         // Step 1. Aggregate by abstraction level
         TT2HFLTSMHTWOWAMultiLevelOperator tt2HFLTSMHTWOWAMultiLevelOperator = new TT2HFLTSMHTWOWAMultiLevelOperator();
-        List<ArrayList<ArrayList<TT2HFLTS>>> allByLevel = tt2HFLTSMHTWOWAMultiLevelOperator.aggregateByAbstractionLevel(model.getCriteria(), model.getAbstractionLevels(), all, targetScaleSize, model.getCriteriaWeightsPerGroup());
 
-        List<ArrayList<ArrayList<TT2HFLTS>>> allByExpert = tt2HFLTSMHTWOWAMultiLevelOperator.transposeByAbstractionLevel(model.getAbstractionLevels().size(), model.getAlternatives().size(), model.getExperts().size(), allByLevel);
+        List<ArrayList<ArrayList<TT2HFLTS>>> allByLevel = tt2HFLTSMHTWOWAMultiLevelOperator.aggregateByAbstractionLevel(
+                model.getCriteria(), model.getAbstractionLevels(), all, targetScaleSize,
+                model.getCriteriaWeightsPerGroup());
+
+        List<ArrayList<ArrayList<TT2HFLTS>>> allByExpert =
+                tt2HFLTSMHTWOWAMultiLevelOperator.transposeByAbstractionLevel(model.getAbstractionLevels().size(),
+                        model.getAlternatives().size(), model.getExperts().size(), allByLevel);
 
         int numExperts = model.getExpertWeightsRule().values().size();
         float[] a = new float[numExperts];
@@ -132,14 +139,15 @@ private void processMultiLevelAdvancedSample(File inputFile, File outputDirector
         if (numExperts > 1) {
             a[1] = 1 - curMax;
         }
-        List<ArrayList<TT2HFLTS>> altToLevel = tt2HFLTSMHTWOWAMultiLevelOperator.aggregateByExpert(model.getAbstractionLevels().size(), model.getAlternatives().size(), 7, allByExpert, a);
+        List<ArrayList<TT2HFLTS>> altToLevel = tt2HFLTSMHTWOWAMultiLevelOperator.aggregateByExpert(
+                model.getAbstractionLevels().size(), model.getAlternatives().size(), targetScaleSize, allByExpert, a);
 
-        List<TT2HFLTS> altVec = tt2HFLTSMHTWOWAMultiLevelOperator.aggregateFinalAltEst(7, altToLevel);
+        List<TT2HFLTS> altVec = tt2HFLTSMHTWOWAMultiLevelOperator.aggregateFinalAltEst(targetScaleSize, altToLevel);
 
         // all below is just processing
 
         // saving to file
-        TTJSONOutputModel res = TTUtils.prepareAllResultsForJSON(altVec, model, 7);
+        TTJSONOutputModel res = TTUtils.prepareAllResultsForJSON(altVec, model, targetScaleSize);
         Path outputJSONFilePath = Paths.get(outputDirectory.toString(), "result.json");
         TTJSONUtils.getInstance().writeResultToJSON(outputJSONFilePath, res);
 
@@ -157,14 +165,8 @@ private void processMultiLevelAdvancedSample(File inputFile, File outputDirector
     }
 
     public void go(File inputFile, File outputDirectory) throws IOException {
-        processSimpleCase(outputDirectory);
-
-        processNumericSample();
-
-        processweightsGenerationSample();
-
-        processMHTWASample();
-
+//        processSimpleCase(outputDirectory);
+//        processNumericSample();
         processMultiLevelAdvancedSample(inputFile, outputDirectory);
     }
 }
diff --git a/src/main/java/dss/lingvo/utils/TTUtils.java b/src/main/java/dss/lingvo/utils/TTUtils.java
@@ -300,6 +300,14 @@ public static List<ArrayList<ArrayList<TT2HFLTS>>> getAllEstimationsFromMultiLev
 
                             float numericEstimation = valueToRemember / (averages.get(critEst.getCriteriaID()).floatValue());
 
+                            if (!criterion.isBenefit()) {
+                                assert numericEstimation >= 0. && numericEstimation <= 1: "should be already a normalized value";
+
+                                // hereinafter all criteria are considered as benefit, therefore need to
+                                // reverse cost criteria by subtracting them from one
+                                numericEstimation = 1 - numericEstimation;
+                            }
+
                             List<Float> fSet = translator.getFuzzySetForNumericEstimation(numericEstimation, targetScaleSize);
                             float resTranslation = translator.getTranslationFromFuzzySet(fSet);
                             TTTuple resTuple = translator.getTTupleForNumericTranslation(resTranslation, targetScaleSize);