// local
public static Pipeline GetPipeline(TaskKind task, IDataView data, string label)
{
var mlContext = new MLContext();
var availableTransforms = TransformInferenceApi.InferTransforms(mlContext, data, label);
var availableTrainers = RecipeInference.AllowedTrainers(mlContext, task, 1);
var pipeline = new InferredPipeline(availableTransforms, availableTrainers.First(), mlContext);
return(pipeline.ToPipeline());
}
public static IEstimator <ITransformer> InferTransforms(this TransformsCatalog catalog, IDataView data, string label)
{
var mlContext = new MLContext();
var suggestedTransforms = TransformInferenceApi.InferTransforms(mlContext, data, label);
var estimators = suggestedTransforms.Select(s => s.Estimator);
var pipeline = new EstimatorChain <ITransformer>();
foreach (var estimator in estimators)
{
pipeline = pipeline.Append(estimator);
}
return(pipeline);
}
private void IteratePipelinesAndFit()
{
var stopwatch = Stopwatch.StartNew();
var transforms = TransformInferenceApi.InferTransforms(_mlContext, _trainData, _label, _puproseOverrides);
var availableTrainers = RecipeInference.AllowedTrainers(_mlContext, _task, _settings.StoppingCriteria.MaxIterations);
do
{
// get next pipeline
var pipeline = PipelineSuggester.GetNextPipeline(_history, transforms, availableTrainers, _optimizingMetricInfo.IsMaximizing);
// break if no candidates returned, means no valid pipeline available
if (pipeline == null)
{
break;
}
// evaluate pipeline
ProcessPipeline(pipeline);
} while (_history.Count < _settings.StoppingCriteria.MaxIterations &&
stopwatch.Elapsed.TotalMinutes < _settings.StoppingCriteria.TimeOutInMinutes);
}
