@Test
public void prune_givenValidationDataSameAsTrainingData_expect100Percent()
throws FileNotFoundException, UnsupportedEncodingException {
Model model = data.getMushroomModel();
List<Sample> trainingSet = data.getAllMushroomSamples();
List<Prediction> originalPrediction = predictor.predict(model, trainingSet);
Double originalAccuracy = accuracy.evaluate(originalPrediction, model.getTargetAttribute());
List<Rule> prunedRules = pruner.pruneRepeatedly(model, trainingSet);
List<Prediction> prunedPrediction = predictor.predict(prunedRules, trainingSet);
Double prunedAccuracy = accuracy.evaluate(prunedPrediction, model.getTargetAttribute());
log.debug(
"Accuracy before pruning = {}, Accuracy after pruning = {}",
originalAccuracy,
prunedAccuracy);
assertThat(originalAccuracy, is(prunedAccuracy));
}
@Test
public void prune_given25PercentValidationSet_expectNoDecreaseInAccuracy()
throws FileNotFoundException, UnsupportedEncodingException {
data.loadData(25.0);
Model model = data.getMushroomModel();
List<Sample> validationSet = data.getValidationSet();
List<Prediction> originalPrediction = predictor.predict(model, validationSet);
Double originalAccuracy = accuracy.evaluate(originalPrediction, model.getTargetAttribute());
List<Rule> prunedRules = pruner.pruneRepeatedly(model, validationSet);
List<Prediction> prunedPrediction = predictor.predict(prunedRules, validationSet);
Double prunedAccuracy = accuracy.evaluate(prunedPrediction, model.getTargetAttribute());
log.debug(
"Accuracy before pruning = {}%, Accuracy after pruning = {}%",
originalAccuracy, prunedAccuracy);
assertThat(prunedAccuracy, Matchers.greaterThanOrEqualTo(originalAccuracy));
}