static void detectAnomalies()
{
var anomalyDetectionModel = Serializer.Load <SupportVectorMachine <Gaussian> >(AnomalyDetectionModelFileName);
var anomalyClassificationModel = Serializer.Load <SupportVectorMachine <Gaussian> >(AnomalyClassificationModelFileName);
MLService service = new MLService();
SegmentService serviceService = new SegmentService();
var chains = service.GetAllSegments_SegmentChain7_AnomalySpreaded();
int numberOfAnomalies = 0;
int numberOfPotholes = 0;
foreach (var chain in chains)
{
double[] inputs = new double[] { chain.MaxSpeed.Value, chain.MinSpeed.Value, chain.SpeedRange.Value, chain.TotalTime.Value, chain.MinTime.Value, chain.MaxTime.Value, chain.TimeRange.Value, chain.MaxSpeedVar.Value, chain.MinSpeedVar.Value, chain.SpeedVarRange.Value, chain.MaxTimeVar.Value, chain.MinTimeVar.Value, chain.TimeVarRange.Value };
bool isContainAnomaly = anomalyDetectionModel.Decide(inputs);
bool isPothole = false;
if (isContainAnomaly)
{
numberOfAnomalies++;
isPothole = anomalyClassificationModel.Decide(inputs);
if (isPothole)
{
numberOfPotholes++;
}
SegmentChain sc = new SegmentChain()
{
Segment1Id = chain.Segment1Id,
Segment2Id = chain.Segment2Id,
Segment3Id = chain.Segment3Id,
Segment4Id = chain.Segment4Id,
Segment5Id = chain.Segment5Id,
Segment6Id = chain.Segment6Id,
Segment7Id = chain.Segment7Id,
PredictedAnomalyType = (short)(isContainAnomaly? (isPothole?2:1):0),
CreationDate = DateTime.Now
};
serviceService.CreateSegmentChain(sc);
}
}
}
static void trainAnomalyDetectionModel()
{
MLService service = new MLService();
//var chain = service.GetAllSegments_SegmentChain3();
//var chain = service.GetAllSegments_SegmentChain4();
//var chain = service.GetAllSegments_SegmentChain5();
//var chain = service.GetAllSegments_SegmentChain6();
//var chain = service.GetAllSegments_SegmentChain4_HighSampleCount();
//var chain = service.GetAllSegments_SegmentChain5_Top28();
//var chain = service.GetAllSegments_SpeedDifference_SegmentChain5_Minimized();
//var chain = service.GetAllSegments_SegmentChain6_Rounded();
//var chain = service.GetAllSegments_SegmentChain6_AnomalySpreaded();
//var chain = service.GetAllSegments_SegmentChain6_AnomalySpreaded();
var chain = service.GetAllSegments_SegmentChain7_AnomalySpreaded();
// var chain = service.GetAllSegments_SegmentChain7_AnomalySpreaded().Where(t => t.AnomalyType == 1 || t.AnomalyType == 2);
//var chain = service.GetAllSegments_SpeedDifference_SegmentChain5();
//var chain = service.GetAllSegments_HighSamplesCount_SegmentChain5();
//double[][] test = chain1.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value, sc.speedAvg5.Value, sc.speedVar1.Value, sc.speedVar2.Value, sc.speedVar3.Value, sc.speedVar4.Value, sc.speedVar5.Value, sc.AvgSpentTime1.Value, sc.AvgSpentTime2.Value, sc.AvgSpentTime3.Value, sc.AvgSpentTime4.Value, sc.AvgSpentTime5.Value, sc.VarSpentTime1.Value, sc.VarSpentTime2.Value, sc.VarSpentTime3.Value, sc.VarSpentTime4.Value, sc.VarSpentTime5.Value, }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedVar1.Value, sc.speedVar2.Value, sc.speedVar3.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedVar1.Value, sc.speedVar2.Value, sc.speedVar3.Value,sc.AvgSpentTime1.Value, sc.AvgSpentTime2.Value, sc.AvgSpentTime3.Value, sc.VarSpentTime1.Value, sc.VarSpentTime2.Value, sc.VarSpentTime3.Value}).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value, sc.speedAvg5.Value, sc.speedVar1.Value, sc.speedVar2.Value, sc.speedVar3.Value, sc.speedVar4.Value, sc.speedVar5.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedVar1.Value, sc.speedVar2.Value, sc.speedVar3.Value, sc.AvgSpentTime1.Value, sc.AvgSpentTime2.Value, sc.AvgSpentTime3.Value, sc.VarSpentTime1.Value, sc.VarSpentTime2.Value, sc.VarSpentTime3.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value, sc.speedAvg5.Value, sc.AvgSpentTime4.Value}).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value, sc.speedAvg5.Value, sc.speedVar1.Value, sc.speedVar2.Value, sc.speedVar3.Value, sc.speedVar4.Value, sc.speedVar5.Value,sc.AvgSpentTime1.Value, sc.AvgSpentTime2.Value, sc.AvgSpentTime3.Value, sc.AvgSpentTime4.Value, sc.AvgSpentTime5.Value, sc.VarSpentTime1.Value, sc.VarSpentTime2.Value, sc.VarSpentTime3.Value, sc.VarSpentTime4.Value, sc.VarSpentTime5.Value, }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value, sc.speedAvg5.Value, sc.speedAvg6.Value, sc.speedVar1.Value, sc.speedVar2.Value, sc.speedVar3.Value, sc.speedVar4.Value, sc.speedVar5.Value, sc.speedVar6.Value, sc.AvgSpentTime1.Value, sc.AvgSpentTime2.Value, sc.AvgSpentTime3.Value, sc.AvgSpentTime4.Value, sc.AvgSpentTime5.Value, sc.AvgSpentTime6.Value, sc.VarSpentTime1.Value, sc.VarSpentTime2.Value, sc.VarSpentTime3.Value, sc.VarSpentTime4.Value, sc.VarSpentTime5.Value, sc.VarSpentTime6.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value, sc.speedVar1.Value, sc.speedVar2.Value, sc.speedVar3.Value, sc.speedVar4.Value, sc.AvgSpentTime1.Value, sc.AvgSpentTime2.Value, sc.AvgSpentTime3.Value, sc.VarSpentTime1.Value, sc.VarSpentTime2.Value, sc.VarSpentTime3.Value }).ToArray();
// double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value, sc.speedVar1.Value, sc.speedVar2.Value, sc.speedVar3.Value, sc.speedVar4.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value, sc.speedAvg5.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value, sc.speedAvg5.Value, sc.AvgSpentTime1.Value, sc.AvgSpentTime2.Value, sc.AvgSpentTime3.Value, sc.AvgSpentTime4.Value, sc.AvgSpentTime5.Value}).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.AvgSpentTime1.Value, sc.AvgSpentTime2.Value, sc.AvgSpentTime3.Value, sc.AvgSpentTime4.Value, sc.AvgSpentTime5.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.speedAvg1.Value, sc.speedAvg2.Value, sc.speedAvg3.Value, sc.speedAvg4.Value, sc.speedAvg5.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.SpeedDiff1.Value, sc.SpeedDiff2.Value, sc.SpeedDiff3.Value, sc.SpeedDiff4.Value,sc.AvgSpentTime1.Value, sc.AvgSpentTime2.Value, sc.AvgSpentTime3.Value, sc.AvgSpentTime4.Value, sc.AvgSpentTime5.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.SpeedDiff1.Value, sc.SpeedDiff2.Value, sc.SpeedDiff3.Value, sc.SpeedDiff4.Value,sc.AvgSpentTime1.Value, sc.AvgSpentTime2.Value, sc.AvgSpentTime3.Value, sc.AvgSpentTime4.Value, sc.AvgSpentTime5.Value }).ToArray();
//double[][] inputs = chain.Select(sc => new double[] { sc.SpeedDiff1.Value, sc.SpeedDiff2.Value,sc.AvgTime.Value,sc.speedAvg4.Value }).ToArray();
double[][] inputs = chain.Select(sc => new double[] { sc.MaxSpeed.Value, sc.MinSpeed.Value, sc.SpeedRange.Value, sc.TotalTime.Value, sc.MinTime.Value, sc.MaxTime.Value, sc.TimeRange.Value, sc.MaxSpeedVar.Value, sc.MinSpeedVar.Value, sc.SpeedVarRange.Value, sc.MaxTimeVar.Value, sc.MinTimeVar.Value, sc.TimeVarRange.Value }).ToArray(); //
//double[][] inputs = chain.Select(sc => new double[] { sc.TotalTime.Value, sc.MinTime.Value, sc.MaxTime.Value, sc.TimeRange.Value, sc.MaxSpeedVar.Value, sc.MinSpeedVar.Value, sc.SpeedVarRange.Value }).ToArray(); //
int[] outputs = chain.Select(sc => sc.ContainAnomaly).ToArray();
//int[] outputs = chain.Select(sc => sc.AnomalyType).ToArray();
//double[][] inputs0 = chain.Where(sc => sc.ContainAnomaly == 0).Select(sc => new double[] { sc.SpeedDiff1.Value, sc.SpeedDiff2.Value }).ToArray();
//int[] outputs0 = chain.Where(sc => sc.ContainAnomaly == 0).Select(sc => sc.ContainAnomaly).ToArray();
//double[][] inputs1 = chain.Where(sc => sc.ContainAnomaly == 1).Select(sc => new double[] { sc.SpeedDiff1.Value, sc.SpeedDiff2.Value }).ToArray();
//int[] outputs1 = chain.Where(sc => sc.ContainAnomaly == 1).Select(sc => sc.ContainAnomaly).ToArray();
//double[][] inputs2 = chain.Where(sc => sc.ContainAnomaly == 0).Select(sc => new double[] { sc.SpeedDiff1.Value, sc.AvgTime.Value }).ToArray();
//int[] outputs2 = chain.Where(sc => sc.ContainAnomaly == 0).Select(sc => sc.ContainAnomaly).ToArray();
//double[][] inputs3= chain.Where(sc => sc.ContainAnomaly == 1).Select(sc => new double[] { sc.SpeedDiff1.Value, sc.AvgTime.Value }).ToArray();
//int[] outputs3= chain.Where(sc => sc.ContainAnomaly == 1).Select(sc => sc.ContainAnomaly).ToArray();
//SVM_crossValidation(inputs, outputs);
SVM_TrainAnomalyDetectionModel(inputs, outputs);
//SVM_crossValidation_AnomalyDetection(inputs, outputs);
//DecisionTree_crossValidation(inputs, outputs);
//GridSearch(inputs, outputs);
//var pca = new PrincipalComponentAnalysis()
//{
// Method = PrincipalComponentMethod.Center,
// Whiten = true
//};
//// Now we can learn the linear projection from the data
//MultivariateLinearRegression transform = pca.Learn(inputs);
//pca.NumberOfOutputs = 3;
//// Finally, we can project all the data
//double[][] output1 = pca.Transform(inputs);
//SVM_crossValidation(inputs, outputs);
//// Or just its first components by setting
//// NumberOfOutputs to the desired components:
//// And then calling transform again:
//double[][] output2 = pca.Transform(inputs);
//// We can also limit to 80% of explained variance:
//pca.ExplainedVariance = 0.8;
//// And then call transform again:
//double[][] output3 = pca.Transform(inputs);
//// Create a new K-Means algorithm
//KMeans kmeans = new KMeans(k: 2);
////kmeans.Centroids = new double [2][];
////kmeans.Centroids[0] = new double[] {-2.76384235537843, 0.608515794829327 };
////kmeans.Centroids[1] = new double[] {-0.579246221664498,0.121793585323178};
//// Compute and retrieve the data centroids
//var clusters = kmeans.Learn(inputs);
//// Use the centroids to parition all the data
//int[] labels = clusters.Decide(inputs);
//var zeros = labels.Where(l => l == 0).ToList();
//var ones = labels.Where(l => l == 1).ToList();
//var twos = labels.Where(l => l == 2).ToList();
//var threes = labels.Where(l => l == 3).ToList();
//int truleyLabeledAnomalies = 0;
//for (int i = 0; i < chain.Count; i++)
//{
// if (chain.ElementAt(i).ContainAnomaly == 1 && labels.ElementAt(i) == 1)
// {
// truleyLabeledAnomalies += 1;
// }
//}
//Console.WriteLine(truleyLabeledAnomalies);
//var smo = new SequentialMinimalOptimization<Linear>()
//{
// Complexity = 0.000001,
// Kernel = new Linear()
//};
//// Use the algorithm to learn the svm
//var svm = smo.Learn(inputs, outputs);
//// bool pred = svm.Decide(new double[] { 2.49922336915495, 2.38719486132452, 2.59047776529037, 1.94322311520282, 2.32949533880782, 1.53491343472098, 0.0611354785139238, 0.848698199040911, 0.85367931112144, 0.887108397987341, 2, 2.66666666666667, 2, 2.54545454545455, 3, 1.33333333333333, 0, 0.672727272727273, 0.75, 0 });
////bool pred = svm.Decide(new double[] { 3.6668832456934,2.72565059380599,3.4617494242872,4.15125345320445,1.07290177209667,4.02676082352871,0.157250607457867,0.00245768485574516,8.51233980164814,1.2701659036119,2,2,2,10,2.33333333333333,0,0 ,0,128,0.333333333333334});
////bool pred = svm.Decide(new double[] {0.342804868030116,0.971802788698728,3.87391319346965, 0.0486282126059593,2.94649397465733,0.547271634138705, 0 ,4.34382476637454 , 0.00508098235164243,1.13588971473715,2,2,1.75 ,28 ,1 , 1 , 0 , 0.25 , 1352 , 0 });
//bool[] prediction = svm.Decide(test);
//// Compute the classification error between the expected
//// values and the values actually predicted by the machine:
//double error = new AccuracyLoss(outputs).Loss(prediction);
//Console.WriteLine("Error: " + error);
//Show results on screen
//ScatterplotBox.Show("not anomaly speed", inputs0, outputs0);
//ScatterplotBox.Show("anomaly speed", inputs1, outputs1);
//ScatterplotBox.Show("not anomaly time", inputs2, outputs2);
//ScatterplotBox.Show("anomaly time ", inputs3, outputs3);
//ScatterplotBox.Show("SVM results", inputs, prediction.ToZeroOne());
Console.ReadKey();
}