public static int GetKNNVersion(List <List <string> > Dataset, KNNVersion Version)
{
List <string> DistinctClasses = new List <string>();
foreach (var Item in Dataset)
{
DistinctClasses.Add(Item.Last());
}
// Guarda todas as classes sem repeticao
DistinctClasses = DistinctClasses.Distinct().ToList();
int P = DistinctClasses.Count; //Quantidade de Classes
int Q = Dataset.Count; //Quantidade de linhas
int M = P % 2 == 0 ? P + 1 : P;
switch (Version)
{
case KNNVersion.OneNN:
return(1);
case KNNVersion.M2NN:
{
return(M + 2);
}
case KNNVersion.M10NN:
{
return(M * 10 + 1);
}
case KNNVersion.QNN:
return((Q / 2) % 2 == 0 ? Q / 2 + 1 : Q / 2);
}
return(-1);
}
public static void TestKNN(string path, KNNVersion KNNVersion)
{
List <List <string> > TrainingSet, TestingSet, DataSet;
List <float> AccuracyRatings = new List <float>();
int StartIndex = 0, KFold = 10, K = 1;
KNN kt = new KNN();
DataSet = kt.LoadCSVData(path);
//GET KNN VERSION
K = GetKNNVersion(DataSet, KNNVersion);
List <float> listOfErroAmostral = new List <float>();
// Calcula dataset de traino e teste
int etapa = 1;
while (StartIndex < DataSet.Count)
{
StartIndex = kt.PrepareDataset(StartIndex, KFold, ref DataSet, out TestingSet, out TrainingSet);
//Gera previsoes
List <string> predictions = new List <string>();
// Calcula label da primeira linha do set de teste
for (int x = 0; x < TestingSet.Count; x++)
{
List <LineDistance> neighbors = kt.GetNeighbors(TrainingSet, TestingSet[x], K);
string result = kt.getResponses(neighbors);
predictions.Add(result);
Console.Write("\rKFold {0} / {1}... {2}%", etapa, KFold, Math.Round(((float)(x + 1) / (float)TestingSet.Count) * 100));
}
// Calcula precisao da label calculada
float acc = kt.getAccuracy(TestingSet, predictions);
AccuracyRatings.Add(acc);
// Exibe precisao media
Console.WriteLine("\nPrecisao Media: " + AccuracyRatings.Average() + "%");
// Guarda erro amostral da linha
listOfErroAmostral.Add(CrossValidation.erroAmostral(TestingSet, predictions));
CrossValidation.prepareConfusionMatrix(DataSet, TestingSet, predictions);
etapa++;
}
/*if (CrossValidation.binaryConfusionMatrix.Count > 0) //Se for matriz binaria ira salvar os dados aqui
* {
* string fileName = path.Split('/').Last().Split('-')[0];
* FileSystem.SaveFileContents(CrossValidation.GeraMatrizBinaria(), @"../../Raw Data/Normalized/output/" + fileName + "/", fileName + "-Matriz-Binaria-Confusao-" + KNNVersion.ToString() +".txt");
* }
* if(CrossValidation.multiClassConfusionMatrix.Count > 0) //Se for matriz multi-classe irá salvar aqui
* {
* string fileName = path.Split('/').Last().Split('-')[0];
* FileSystem.SaveFileContents(CrossValidation.GeraMatrizMultiClasse(), @"../../Raw Data/Normalized/output/" + fileName + "/", fileName + "-Matriz-MultiClasse-Confusao-" + KNNVersion.ToString() + ".txt");
* }
*
* CrossValidation.erroDeValidacaoCruzada(listOfErroAmostral);*///Verifica a taxa de erro da validacao cruzada
}
