private static List <double[]> SolveWithLinearRegression(DataSet ds, DataSet dt, LinearModel model)
{
LLSGradientDescent gd = new LLSGradientDescent();
//Calling the Train method of the trainer with the desired parameters
//n, ds, learningRate: .3, numberOfEpochs: 200, shuffle: false, debug: n.Debug, nestrov:false, momentum:0.9, resilient: false, resilientUpdateAccelerationRate: 0.3,
//resilientUpdateSlowDownRate: 0.1, regularization: AA1_MLP.Enums.Regularizations.L2, regularizationRate: 0.001, validationSet: dt, batchSize: 7
LinearLeastSquaresParams passedParams = new LinearLeastSquaresParams {
model = model
};
passedParams.trainingSet = ds;
passedParams.learningRate = 1;
passedParams.numOfIterations = 3000;
passedParams.shuffle = false;
passedParams.debug = false;
passedParams.regularizationRate = 0.01;
passedParams.regularizationType = Regularizations.None;
passedParams.validationSet = dt;
passedParams.degree = 30;
var learningCurve = gd.Train(passedParams);
return(learningCurve);
}
/// <summary>
/// For outputting the LLS solution results
/// </summary>
/// <param name="trainsetpath"></param>
/// <param name="testsetpath"></param>
private static void CupTestingLLS(string trainsetpath, string testsetpath)
{
CupDataManager dm = new CupDataManager();
DataSet trainset = dm.LoadData(trainsetpath, 10, 2, standardize: true);
DataSet testset = dm.LoadData(testsetpath, 10, 2, standardize: true);
LinearModel model = new LinearModel {
bias = true
};
//**trying SVD
LinearLeastSquaresParams passedParams = new LinearLeastSquaresParams {
model = model, numOfIterations = 5000, learningRate = 0.1, degree = 1
};
passedParams.trainingSet = trainset;
passedParams.validationSet = testset;
Console.WriteLine("SVD solution:");
var svdlearningCurve = new LLSSVD().Train(passedParams);
Console.WriteLine("Normal Equations Solution:");
var normallearningCurve = new LLSNormal().Train(passedParams);
Console.WriteLine("GD Solution:");
var gdlearningCurve = new LLSGradientDescent().Train(passedParams);
}
