Package etc.aloe.data

Examples of etc.aloe.data.EvaluationReport

63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
            //Run cross validation
            crossValidationController.setSegmentSet(segments);
            crossValidationController.run();

            EvaluationReport evalReport = crossValidationController.getEvaluationReport();
            if (evalReport != null) {
                System.out.println("== Saving Results of Cross Validation ==");
               
                saveEvaluationReport(evalReport, options.outputEvaluationReportFile);
               
                System.out.println("Aggregated cross-validation report:");
                System.out.println(evalReport);
                System.out.println("---------");

                if (options.makeROC) {
                    options.outputROCDir.mkdirs();
                   
                    for (ROC roc : evalReport.getROCs()) {
                        String fileName = roc.getName() + FileNames.ROC_SUFFIX;
                        File outputFile = new File(options.outputROCDir, fileName);

                        saveROC(roc, outputFile);
                    }
                }
               
                if (options.outputTests) {
                    options.outputTestsDir.mkdirs();
                    List<SegmentSet> testSets = evalReport.getTestSets();
                    List<String> testSetNames = evalReport.getTestSetNames();
                   
                    SegmentSet combined = new SegmentSet();
                   
                    for (int i = 0; i < testSets.size(); i++) {
                        String fileName = testSetNames.get(i) + FileNames.TEST_DATA_SUFFIX;
View Full Code Here
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
            labelingController.setSegmentSet(segments);
            labelingController.setFeatureSpecification(spec);
            labelingController.run();

            //Get the outputs
            EvaluationReport evalReport = labelingController.getEvaluationReport();

            System.out.println("== Saving Output ==");

            saveEvaluationReport(evalReport, options.outputEvaluationReportFile);
            if (options.makeROC) {
                ROC roc = evalReport.getROCs().get(0);
                saveROC(roc, options.outputROCFile);
            }
           
            if (options.outputFeatureValues) {
                Instances featureValues = labelingController.getFeatureValues();
View Full Code Here
77
78
79
80
81
82
83
84
        //Map back onto messages
        LabelMapping mapping = getMappingImpl();
        mapping.map(predictions, segmentSet);

        //Evaluate the model on labeled examples
        this.evaluationReport = new EvaluationReport("Unlabeled Data", falsePositiveCost, falseNegativeCost);
        this.evaluationReport.addPredictions(predictions);
    }
View Full Code Here
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
            System.out.println("Randomizing and stratifying segments.");
            CrossValidationPrep<Segment> validationPrep = new CrossValidationPrep<Segment>();
            validationPrep.randomize(segmentSet.getSegments());
            segmentSet.setSegments(validationPrep.stratify(segmentSet.getSegments(), folds));

            evaluationReport = new EvaluationReport(this.folds + " Cross Validation", falsePositiveCost, falseNegativeCost);
            for (int foldIndex = 0; foldIndex < this.folds; foldIndex++) {
                System.out.println("- Starting fold " + (foldIndex + 1));
                //Split the data
                CrossValidationSplit<Segment> split = new CrossValidationSplit<Segment>();

                System.out.println("- Splitting out training set");
                SegmentSet trainingSegments = new SegmentSet();
                trainingSegments.setSegments(split.getTrainingForFold(segmentSet.getSegments(), foldIndex, this.folds));
                if (getBalancingImpl() != null) {
                    trainingSegments = getBalancingImpl().balance(trainingSegments);
                }

                System.out.println("- Extracting basic features from training set");
                ExampleSet basicTrainingExamples = trainingSegments.getBasicExamples();
                trainingSegments = null;

                FeatureGeneration generation = getFeatureGenerationImpl();
                System.out.println("- Generating features");
                FeatureSpecification spec = generation.generateFeatures(basicTrainingExamples);

                FeatureExtraction extraction = getFeatureExtractionImpl();
                System.out.println("- Extracting features from training set");
                ExampleSet trainingSet = extraction.extractFeatures(basicTrainingExamples, spec);
                basicTrainingExamples = null;

                Training training = getTrainingImpl();
                Model model = training.train(trainingSet);
                trainingSet = null;
               
                System.out.println("- Splitting out test set");
                SegmentSet testingSegments = new SegmentSet();
                testingSegments.setSegments(split.getTestingForFold(segmentSet.getSegments(), foldIndex, this.folds));
                if (getBalancingImpl() != null && balanceTestSet) {
                    testingSegments = getBalancingImpl().balance(testingSegments);
                }

                System.out.println("- Extracting basic features from test set");
                ExampleSet basicTestingExamples = testingSegments.getBasicExamples();

                System.out.println("- Extracting features from test set");
                ExampleSet testingSet = extraction.extractFeatures(basicTestingExamples, spec);
                basicTestingExamples = null;

                Predictions predictions = model.getPredictions(testingSet);
                EvaluationReport report = new EvaluationReport("Fold " + (foldIndex + 1), falsePositiveCost, falseNegativeCost);
                report.addPredictions(predictions);
               
                LabelMapping mapping = getMappingImpl();
                mapping.map(predictions, testingSegments);
                report.addLabeledTestData(testingSegments);
               
                evaluationReport.addPartial(report);
                int numCorrect = report.getTrueNegativeCount() + report.getTruePositiveCount();
                System.out.println("- Fold " + (foldIndex + 1) + " completed (" + numCorrect + "/" + testingSet.size() + " correct).");
                System.out.println();
            }
        } else {
            System.out.println("== Skipping Cross Validation ==");
View Full Code Here
TOP

Related Classes of etc.aloe.data.EvaluationReport

  • etc.aloe.AloeLabel
  • etc.aloe.AloeTrain
  • etc.aloe.controllers.CrossValidationController
  • etc.aloe.controllers.LabelingController
  • java.io.PrintStream
    • Copyright © 2018 www.massapicom. All rights reserved.
    All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.