public Insert2 GetModelStatus(ProjectModelId projectModelId)
{
var getRequest = PredictionService.Trainedmodels.Get(projectModelId.ProjectNumber, projectModelId.ModelId);
Insert2 response = getRequest.Execute();
return(response);
}
public void Run()
{
Insert2 trainingStatus = m_predictionFramework.GetModelStatus(ProjectModelId);
Console.WriteLine("Training status: {0}", trainingStatus.TrainingStatus);
ToConsole(trainingStatus.ModelInfo);
}
private static void ToConsole(Insert2.ModelInfoData modelInfoData)
{
if (modelInfoData != null)
{
Console.WriteLine("ModelType: {0}", modelInfoData.ModelType);
Console.WriteLine("NumberInstances: {0}", modelInfoData.NumberInstances);
Console.WriteLine("MeanSquaredError: {0}", modelInfoData.MeanSquaredError);
}
}
public Insert2 TrainRegressionModel(ProjectModelId projectModelId, string csvDataPathInStorage)
{
Insert insertBody = new Insert
{
StorageDataLocation = csvDataPathInStorage,
Id = projectModelId.ModelId,
ModelType = "REGRESSION",
};
TrainedmodelsResource.InsertRequest insertRequest = PredictionService.Trainedmodels.Insert(insertBody, projectModelId.ProjectNumber);
Insert2 insertResponse = insertRequest.Execute();
return(insertResponse);
}
public void Run()
{
Console.WriteLine("Model '{0}' deleted with response '{1}'.", ProjectModelId.ModelId, m_predictionFramework.DeleteTrainedModel(ProjectModelId));
Insert2 insertResponse = m_predictionFramework.TrainRegressionModel(ProjectModelId, m_storageData);
Console.WriteLine("Inserted the training data for the model.");
// Wait until the training is complete
bool trainingRunning = true;
while (trainingRunning)
{
Console.WriteLine("Getting a new training progress status...");
var getResponse = m_predictionFramework.GetModelStatus(ProjectModelId);
Console.WriteLine("Got a new training progress status: {0}", getResponse.TrainingStatus);
switch (getResponse.TrainingStatus)
{
case "RUNNING":
Console.WriteLine("The model training is still in progress, let us wait for {0} ms.", PROGRESS_WAITING_TIME);
Thread.Sleep(PROGRESS_WAITING_TIME);
break;
case "DONE":
Console.WriteLine("The model has been trained successfully.");
ToConsole(getResponse.ModelInfo);
trainingRunning = false;
break;
case "ERROR: TRAINING JOB NOT FOUND":
throw new Exception("the training job was not found.");
case "ERROR: TOO FEW INSTANCES IN DATASET":
throw new Exception("there are too few instances in the dataset.");
default:
throw new ArgumentException("Unknown status (error): " + getResponse.TrainingStatus);
}
}
}
