public void TestKnearestClassification()
{
var r = new Random();
var clusters = new Dictionary <int, IEnumerable <GenericVector> >();
var samplePoint = new GenericVector(96, 86);
var clus = 1;
for (int i = 50; i < 201; i += 50)
{
var sampleData = new List <GenericVector>();
for (int j = 0; j < 4; j++)
{
sampleData.Add(new GenericVector(r.Next(i - 10, i), r.Next(i - 10, i)));
}
clusters[clus++] = sampleData;
}
var kNearest = new KnearestClassification(clusters, 3);
Assert.Equal(2, kNearest.ClassifyPoint(samplePoint));
}
public string CreateGraph(Stud dataA, Stud dataB, bool kmeans, bool dbscan, bool linearregression,
bool polynomialregression, bool pearsoncorrelation, bool spearmancorrelation, bool knearest, bool naivebayes)
{
var gradedStudents = Students.StudentsGraded;
var classifyungraded = knearest || naivebayes;
if (classifyungraded)
{
var grades = new Dictionary <int, double>();
var clusters = new Dbscan(210, 3,
Students.StudentsGraded.Select(x => x.ToGenericVector(Stud.Attempts, Stud.Class, Stud.FailRatio,
Stud.Fails, Stud.Succeeds, Stud.SuccessRatio, Stud.Grade)));
foreach (var cluster in clusters.DataClusters)
{
grades[cluster.Key] = cluster.Value.Sum(x => x[6]) / cluster.Value.Count();
}
Classification classification = new NaiveBayesClassification(clusters.DataClusters, 50000);
if (knearest)
{
classification = new KnearestClassification(clusters.DataClusters, 5);
}
foreach (var student in Students.StudentsUngraded)
{
var cluster = classification.ClassifyPoint(student.ToGenericVector(Stud.Attempts, Stud.Class,
Stud.FailRatio, Stud.Fails, Stud.Succeeds, Stud.SuccessRatio, Stud.Grade));
student.Grade = (int)grades[cluster];
gradedStudents.Add(student);
}
}
var list = new List <int>();
var a = list
.GroupBy(x => x)
.Select(x => x.OrderBy(y => y))
.Select(x => x.First());
foreach (var student in gradedStudents)
{
student.Filter();
}
var data = new Dataset(gradedStudents.Select(x => x.ToGenericVector(dataA, dataB)));
var highChart = new HighchartsAdapter(Highchart.Scatterplot);
//Dbscan removes outliers, so we have to change are dataset afterwards
if (dbscan)
{
var newData = new List <GenericVector>();
var dBscan = new Dbscan(50, 3, data);
foreach (var cluster in dBscan.DataClusters)
{
newData.AddRange(cluster.Value);
}
data = new Dataset(newData);
highChart.AddClusters(dBscan);
}
if (kmeans)
{
highChart.AddClusters(new Kmeans(4, 100, data));
}
if (linearregression)
{
highChart.AddRegression(new LinearRegression(data.Select(x => x.ToVector2())));
}
if (polynomialregression)
{
highChart.AddRegression(new PolynomialRegression(data.Select(x => x.ToVector2()), 3));
}
if (pearsoncorrelation)
{
highChart.AddCorrelation(new PearsonCorrelation(data.Select(x => x.ToVector2())));
}
if (spearmancorrelation)
{
highChart.AddCorrelation(new SpearmanCorrelation(data.Select(x => x.ToVector2())));
}
highChart.SetDivId("plotkmeans");
highChart.SetTitle($"{dataA} vs {dataB}");
highChart.SetXlabel(dataA.ToString());
highChart.SetYlabel(dataB.ToString());
return(highChart.CreateTemplate());
}