public void SaveToCsvTest()
{
// Setup results
BlockingCollection <TrainingHelper.ParameterResult> results =
new BlockingCollection <TrainingHelper.ParameterResult>();
int logTo = int.Parse(Configuration.Get("logTo"));
int logFrom = int.Parse(Configuration.Get("logFrom"));
for (double cLog = logFrom; cLog <= logTo; cLog++)
{
for (double gammaLog = logFrom; gammaLog <= logTo; gammaLog++)
{
results.Add(new TrainingHelper.ParameterResult()
{
Accuracy = 1, C = Math.Pow(2, cLog),
Gamma = Math.Pow(2, gammaLog)
});
}
}
File.Delete(@"svmDataActual.txt");
TrainingHelper.SaveToCsv(results, "svmDataActual.txt");
FileInfo expectedFile = new FileInfo(@"svmDataExpected.txt");
FileInfo actualFile = new FileInfo(@"svmDataActual.txt");
FileAssert.AreEqual(expectedFile, actualFile);
}
public void SaveToCsvOutOfBoundsTest()
{
// Setup results
BlockingCollection <TrainingHelper.ParameterResult> results =
new BlockingCollection <TrainingHelper.ParameterResult>();
File.Delete(@"svmDataActual.txt");
Assert.Throws <IndexOutOfRangeException>(() => TrainingHelper.SaveToCsv(results, "svmDataActual.txt"));
}
static void Main(string[] args)
{
var mlContext = new MLContext(seed: 0);
// Load data
var data = mlContext.Data.LoadFromTextFile <TyreStint>(DatasetsLocation, ';', true);
// Filtering data
var filtered = mlContext.Data.FilterByCustomPredicate(data, (TyreStint row) => !(row.Reason.Equals("Pit Stop") || row.Reason.Equals("Race Finish")));
var debug = mlContext.Data.CreateEnumerable <TyreStint>(filtered, reuseRowObject: false).Count();
// Divide dataset into training and testing data
var split = mlContext.Data.TrainTestSplit(filtered, testFraction: 0.1);
var trainingData = split.TrainSet;
var testingData = split.TestSet;
// Run AutoML experiment
var experimentTime = 900u;
Console.WriteLine("=============== Training the model ===============");
Console.WriteLine($"Running AutoML regression experiment for {experimentTime} seconds...");
var experimentResult = mlContext.Auto()
.CreateRegressionExperiment(experimentTime)
.Execute(trainingData, testingData,
columnInformation: new ColumnInformation()
{
CategoricalColumnNames = { nameof(TyreStint.Team), nameof(TyreStint.Car), nameof(TyreStint.Driver), nameof(TyreStint.Compound), nameof(TyreStint.Reason) },
NumericColumnNames = { nameof(TyreStint.AirTemperature), nameof(TyreStint.TrackTemperature) },
LabelColumnName = nameof(TyreStint.Distance)
}
);
// Print top models found by AutoML
TrainingHelper.PrintTopModels(experimentResult);
Console.WriteLine("===== Evaluating model's accuracy with test data =====");
var best = experimentResult.BestRun;
var trainedModel = best.Model;
var predictions = trainedModel.Transform(testingData);
var metrics = mlContext.Regression.Evaluate(predictions, labelColumnName: nameof(TyreStint.Distance), scoreColumnName: "Score");
TrainingHelper.PrintRegressionMetrics(best.TrainerName, metrics);
// Run sample predictions
var predictionEngine = mlContext.Model.CreatePredictionEngine <TyreStint, TyreStintPrediction>(trainedModel);
var lh = new TyreStint()
{
Track = "Bahrain International Circuit", TrackLength = 5412f, Team = "Mercedes", Car = "W12", Driver = "Lewis Hamilton", Compound = "C3", AirTemperature = 20.5f, TrackTemperature = 28.3f, Reason = "Pit Stop"
};
var lhPred = predictionEngine.Predict(lh);
var lhLaps = lhPred.Distance / 5412f;
var mv = new TyreStint()
{
Track = "Bahrain International Circuit", TrackLength = 5412f, Team = "Red Bull", Car = "RB16B", Driver = "Max Verstappen", Compound = "C3", AirTemperature = 20.5f, TrackTemperature = 28.3f, Reason = "Pit Stop"
};
var mvPred = predictionEngine.Predict(mv);
var mvLaps = mvPred.Distance / 5412f;
// Printing predictions for top10 grid places
var top10Monaco = new List <Top10Driver>()
{
new Top10Driver()
{
Team = "Ferrari", Car = "SF21", Name = "Charles Leclerc", StartingCompound = "C5"
},
new Top10Driver()
{
Team = "Red Bull", Car = "RB16B", Name = "Max Verstappen", StartingCompound = "C5"
},
new Top10Driver()
{
Team = "Mercedes", Car = "W12", Name = "Valtteri Bottas", StartingCompound = "C5"
},
new Top10Driver()
{
Team = "Ferrari", Car = "SF21", Name = "Carlos Sainz", StartingCompound = "C5"
},
new Top10Driver()
{
Team = "McLaren", Car = "MCL35M", Name = "Lando Norris", StartingCompound = "C5"
},
new Top10Driver()
{
Team = "Toro Rosso / AlphaTauri", Car = "AT02", Name = "Pierre Gasly", StartingCompound = "C5"
},
new Top10Driver()
{
Team = "Mercedes", Car = "W12", Name = "Lewis Hamilton", StartingCompound = "C5"
},
new Top10Driver()
{
Team = "Force India / Racing Point / Aston Martin", Car = "AMR21", Name = "Sebastian Vettel", StartingCompound = "C5"
},
new Top10Driver()
{
Team = "Red Bull", Car = "RB16B", Name = "Sergio Pérez", StartingCompound = "C5"
},
new Top10Driver()
{
Team = "Sauber / Alfa Romeo", Car = "C41", Name = "Antonio Giovinazzi", StartingCompound = "C5"
},
new Top10Driver()
{
Team = "Renault / Alpine", Car = "A521", Name = "Esteban Ocon", StartingCompound = "C5"
},
};
DataHelper.PrintPredictionTable(predictionEngine, "Monaco", 20.0f, 35.0f, top10Monaco);
}
public void OnMapReady(GoogleMap googleMap)
{
trainingHelper = new TrainingHelper(googleMap, CurrentTrainingState);
}
public void FindBestHyperparametersGivenAnswersTest()
{
SVMProblem problem = new SVMProblem();
problem.Add(new SVMNode[]
{
new SVMNode(1, 1),
new SVMNode(2, 10),
new SVMNode(3, 72),
new SVMNode(4, 55),
new SVMNode(5, 1),
}, 1);
problem.Add(new SVMNode[]
{
new SVMNode(1, 1),
new SVMNode(2, 10),
new SVMNode(3, 2),
new SVMNode(4, 95),
new SVMNode(5, 16),
}, 1);
problem.Add(new SVMNode[]
{
new SVMNode(1, 1),
new SVMNode(2, 12),
new SVMNode(3, 13),
new SVMNode(4, 14),
new SVMNode(5, 15),
}, 1);
problem.Add(new SVMNode[]
{
new SVMNode(1, 0),
new SVMNode(2, 13),
new SVMNode(3, 37),
new SVMNode(4, 4),
new SVMNode(5, 18),
}, 0);
problem.Add(new SVMNode[]
{
new SVMNode(1, 0),
new SVMNode(2, 100),
new SVMNode(3, 720),
new SVMNode(4, 550),
new SVMNode(5, 10),
}, 0);
//Setup test SVMParameter
SVMParameter parameter = new SVMParameter
{
Type = SVMType.C_SVC,
Kernel = SVMKernelType.RBF,
C = 2,
Gamma = 1,
Probability = true,
};
var actualParameter = TrainingHelper.FindBestHyperparameters(problem, parameter);
Assert.AreEqual(0.015625d, actualParameter.C);
Assert.AreEqual(0.125d, actualParameter.Gamma);
}
public void GetProblemFromImageModelResultListTest()
{
ushort[,] image1 = new ushort[, ]
{
{ 1, 2, 3, 4, 5 },
{ 6, 2, 8, 9, 10 },
{ 11, 8, 2, 10, 11 },
{ 16, 13, 11, 15, 16 },
{ 21, 2, 1, 1, 25 },
};
ushort[,] image2 = new ushort[, ]
{
{ 1, 2, 3, 4, 5 },
{ 6, 7, 8, 9, 10 },
{ 11, 8, 9, 10, 11 },
{ 16, 13, 14, 15, 16 },
{ 21, 2, 23, 24, 25 },
};
ImageWithResultModel resultImage1 = new ImageWithResultModel();
resultImage1.Image = new UShortArrayAsImage(image1);
resultImage1.Result = 1;
ImageWithResultModel resultImage2 = new ImageWithResultModel();
resultImage2.Image = new UShortArrayAsImage(image2);
resultImage2.Result = 0;
var images = new List <ImageWithResultModel>()
{
resultImage1,
resultImage2
};
File.Delete(@"pca_model-100x100-CC-Train-MassCalc-BI.bin");
PCA pca = TrainingHelper.GetPca(images);
PCA pca2 = TrainingHelper.GetPca(images);
SVMProblem problem = ProblemHandler.GetProblemFromImageModelResultList(images, pca, 10);
//Expected value
SVMProblem realValue = new SVMProblem();
realValue.Add(new SVMNode[]
{
new SVMNode(1, 20.056853498561878),
new SVMNode(2, 13.190302568584602),
new SVMNode(3, -1.0813980605883611),
new SVMNode(4, 0.38976510872122916),
new SVMNode(5, 8.8596355929840787),
new SVMNode(6, -7.3433006502883726),
new SVMNode(7, 10.837768344992746),
new SVMNode(8, 20.626727358988219),
new SVMNode(9, -1.7552480617394235),
new SVMNode(10, 25),
}, 0);
realValue.Add(new SVMNode[]
{
new SVMNode(1, 22.292105243883533),
new SVMNode(2, 11.126898461982794),
new SVMNode(3, -2.3028333386433371),
new SVMNode(4, -6.2077696783291429),
new SVMNode(5, 12.172991455602181),
new SVMNode(6, -4.385545310384054),
new SVMNode(7, 13.837367251719812),
new SVMNode(8, 20.646721554636255),
new SVMNode(9, -1.8000434956830436),
new SVMNode(10, 25),
}, 1);
bool fail = false;
for (int i = 0; i < realValue.Y.Count; i++)
{
// Expected values
var y = realValue.Y[i];
var x = realValue.X[i];
// Actual values
var py = problem.Y[i];
var px = problem.X[i];
for (int j = 0; j < x.Length; j++)
{
if (x[j].Value != px[j].Value)
{
fail = true;
}
}
}
if (!fail)
{
Assert.Pass();
}
else
{
Assert.Fail();
}
}
