/// <summary>
/// Classifies the [features] vector with the trained model.
/// Returns the classification prediction as an integer.
/// Model must first be [Fit] before performing predictions.
/// If model is not [Fit] or the provided feature vector does not match the length of the training data features
/// the function returns NULL.
/// </summary>
/// <param name="features">Features vector that is to be classified</param>
/// <returns>
/// A integer value representing the predicted class of the feature vector.
/// NULL if something goes wrong.
/// </returns>
public int?Predict(FeaturesVector features)
{
if (!this.trained)
{
// DEV NOTE: Not best practice, since the targets provided to the classifier might contain -1 as a class
return(null);
}
if (features.Count != this.featuresVectorLength)
{
Log.Error("NaiveBayesClassifier.Predict", "The feature vector provided is not the same shape as the training data");
// DEV NOTE: Not best practice, since the targets provided to the classifier might contain -1 as a class
return(null);
}
Dictionary <int, double> posteriors = new Dictionary <int, double>();
foreach (int target in this.classes)
{
double sum = 0;
for (int i = 0; i < this.featuresVectorLength; i++)
{
double value = features[i];
double p_likelihood = 0.0;
try
{
p_likelihood = this.likelihood[i][value][target];
}
catch (KeyNotFoundException)
{
// Skipping features that contain values that have not been seen during training
continue;
}
double normalization_factor = this.classes.Select(x => this.likelihood[i][value][x] * this.prior[x]).Sum();
double p_posterior = (p_likelihood * prior[target]) / normalization_factor;
// Computing combined posterior probability for all features
sum += Math.Log(1 - p_posterior) - Math.Log(p_posterior);
}
double posterior = Math.Pow(Math.E, sum);
posterior += 1;
posterior = 1.0 / posterior;
posteriors[target] = posterior;
}
/// Implementation for ArgMax
/// From @"https://stackoverflow.com/questions/2805703/good-way-to-get-the-key-of-the-highest-value-of-a-dictionary-in-c-sharp"
int max = posteriors.Aggregate((l, r) => l.Value > r.Value ? l : r).Key;
//Console.WriteLine(string.Format("spam: {0}, ham: {1}, max: {2}", posteriors[0], posteriors[1], max));
return(max);
}
/// <summary>
/// Function for converting a lentence dataset into a Bag of Words representation to be used for model treining.
/// The text segment (sms) of each data item is split into words. Separation by whitespace and symbols is used.
/// Only words consisting of alphanumeric values are taken into account from each text item.
/// Returns a [DataFeaturesContainer] containing the bag of words as a the [Data] field.
/// </summary>
/// <param name="raw_data">Data to be processed</param>
/// <param name="vocabulary">The vocabulary used for Bag of Words representation</param>
/// <returns></returns>
static DataFeaturesContainer BagOfWordsFromVocabularyAndSMS(List <_DataSetItem> raw_data, HashSet <string> vocabulary)
{
List <FeaturesVector> data = new List <FeaturesVector>();
foreach (var item in raw_data)
{
List <string> sms_vocab = new List <string>(Regex.Split(item.sms, @"[^a-zA-Z0-9]+"));
List <double> tmp = vocabulary.Select(x => sms_vocab.Contains(x) ? 1.0 : 0.0).ToList();
FeaturesVector features = new FeaturesVector(tmp);
data.Add(features);
}
DataFeaturesContainer result = DataFeaturesContainer.Create(data);
return(result);
}