private void CalculateProbabilities()
{
if (clusters != null)
{
foreach (Cluster cluster in clusters)
{
cluster.SetProbability(PredictionFunctions.ProbabiltyOfCluster(cluster, sumSimilarities));
}
}
}
public List <Result> PredictBySelection()
{
if (predictionCouples != null && dailyObservation != null)
{
/// applying similarityFunction To all inputs in predictions Couples
PredictionFunctions.SimilarityFunction(predictionCouples, dailyObservation.ToVector());
List <PredictionCouple> predicted = PredictionFunctions.PickMostProbableCouples(predictionCouples, CLUSTER_WIDTH);
List <Result> results = ConstructResultsBySelection(predicted);
return(results);
}
return(null);
}
/// <summary>
/// Constructs Raw Predictions based on PredictionCouples
/// </summary>
public void PredictByKNN()
{
if (predictionCouples != null && dailyObservation != null)
{
/// applying similarityFunction To all inputs in predictions Couples
PredictionFunctions.SimilarityFunction(predictionCouples, dailyObservation.ToVector());
sumSimilarities = CalculateSumOfSimilarities();
/// clustering the PredictionCouples based on CLUSTER_WIDTH to make the prediction
clusters = PredictionFunctions.Cluster(predictionCouples, CLUSTER_WIDTH);
/// Calculating the probability of each cluster and ordering it by probabilty descending
CalculateProbabilities();
orderedClustersByProba = PredictionFunctions.OrderByProbabiltyDescending(clusters);
/// making the raw predictions and constructing a table containing each cluster with its list of predictions
predictions = PredictionFunctions.Predict(orderedClustersByProba);
/// setting the parameters in order to identify each Prediction
SetParameterKeys();
}
}
