public void SortFeaturesByCountWhenApplyingCountLimit()
{
// arrange
var settings = new CountVectorizerSettings
{
MaxFeatures = 5
};
var target = new CountVectorizer(settings);
var trainingData = GetTrainingData();
target.Fit(trainingData);
var vectorizerWithoutLimit = GetFittedVectorizer(trainingData);
IEnumerable <uint> getTopCounts(IEnumerable <IDictionary <string, uint> > tokensCounts) =>
tokensCounts.SelectMany(kv => kv.Values)
.OrderByDescending(v => v)
.Take((int)settings.MaxFeatures);
var expectedTopCounts = getTopCounts(vectorizerWithoutLimit.Transform(trainingData));
// act
var result = target.Transform(trainingData);
// assert
var resultTopCounts = getTopCounts(result);
Assert.Equal(expectedTopCounts, resultTopCounts);
}
public void CanLimitFeaturesCount()
{
// arrange
var settings = new CountVectorizerSettings
{
MaxFeatures = 5
};
var target = new CountVectorizer(settings);
var trainingData = GetTrainingData();
// act
target.Fit(trainingData);
// assert
Assert.True(target.Vocabulary.Count() == 5);
}
