public static CommonOutputs.TransformOutput FilterByRange(IHostEnvironment env, RangeFilter.Options input)
{
Contracts.CheckValue(env, nameof(env));
var host = env.Register(RangeFilter.LoaderSignature);
host.CheckValue(input, nameof(input));
EntryPointUtils.CheckInputArgs(host, input);
var xf = new RangeFilter(host, input, input.Data);
return(new CommonOutputs.TransformOutput {
Model = new TransformModelImpl(env, xf, input.Data), OutputData = xf
});
}
private FoldResult RunFold(int fold)
{
var host = GetHost();
host.Assert(0 <= fold && fold <= _numFolds);
// REVIEW: Make channels buffered in multi-threaded environments.
using (var ch = host.Start($"Fold {fold}"))
{
ch.Trace("Constructing trainer");
ITrainer trainer = _trainer.CreateComponent(host);
// Train pipe.
var trainFilter = new RangeFilter.Options();
trainFilter.Column = _splitColumn;
trainFilter.Min = (Double)fold / _numFolds;
trainFilter.Max = (Double)(fold + 1) / _numFolds;
trainFilter.Complement = true;
IDataView trainPipe = new RangeFilter(host, trainFilter, _inputDataView);
trainPipe = new OpaqueDataView(trainPipe);
var trainData = _createExamples(host, ch, trainPipe, trainer);
// Test pipe.
var testFilter = new RangeFilter.Options();
testFilter.Column = trainFilter.Column;
testFilter.Min = trainFilter.Min;
testFilter.Max = trainFilter.Max;
ch.Assert(!testFilter.Complement);
IDataView testPipe = new RangeFilter(host, testFilter, _inputDataView);
testPipe = new OpaqueDataView(testPipe);
var testData = _applyTransformsToTestData(host, ch, testPipe, trainData, trainPipe);
// Validation pipe and examples.
RoleMappedData validData = null;
if (_getValidationDataView != null)
{
ch.Assert(_applyTransformsToValidationData != null);
if (!trainer.Info.SupportsValidation)
{
ch.Warning("Trainer does not accept validation dataset.");
}
else
{
ch.Trace("Constructing the validation pipeline");
IDataView validLoader = _getValidationDataView();
var validPipe = ApplyTransformUtils.ApplyAllTransformsToData(host, _inputDataView, validLoader);
validPipe = new OpaqueDataView(validPipe);
validData = _applyTransformsToValidationData(host, ch, validPipe, trainData, trainPipe);
}
}
// Train.
var predictor = TrainUtils.Train(host, ch, trainData, trainer, validData,
_calibrator, _maxCalibrationExamples, _cacheData, _inputPredictor);
// Score.
ch.Trace("Scoring and evaluating");
ch.Assert(_scorer == null || _scorer is ICommandLineComponentFactory, "CrossValidationCommand should only be used from the command line.");
var bindable = ScoreUtils.GetSchemaBindableMapper(host, predictor, scorerFactorySettings: _scorer as ICommandLineComponentFactory);
ch.AssertValue(bindable);
var mapper = bindable.Bind(host, testData.Schema);
var scorerComp = _scorer ?? ScoreUtils.GetScorerComponent(host, mapper);
IDataScorerTransform scorePipe = scorerComp.CreateComponent(host, testData.Data, mapper, trainData.Schema);
// Save per-fold model.
string modelFileName = ConstructPerFoldName(_outputModelFile, fold);
if (modelFileName != null && _loader != null)
{
using (var file = host.CreateOutputFile(modelFileName))
{
var rmd = new RoleMappedData(
CompositeDataLoader.ApplyTransform(host, _loader, null, null,
(e, newSource) => ApplyTransformUtils.ApplyAllTransformsToData(e, trainData.Data, newSource)),
trainData.Schema.GetColumnRoleNames());
TrainUtils.SaveModel(host, ch, file, predictor, rmd, _cmd);
}
}
// Evaluate.
var eval = _evaluator?.CreateComponent(host) ??
EvaluateUtils.GetEvaluator(host, scorePipe.Schema);
// Note that this doesn't require the provided columns to exist (because of the "opt" parameter).
// We don't normally expect the scorer to drop columns, but if it does, we should not require
// all the columns in the test pipeline to still be present.
var dataEval = new RoleMappedData(scorePipe, testData.Schema.GetColumnRoleNames(), opt: true);
var dict = eval.Evaluate(dataEval);
RoleMappedData perInstance = null;
if (_savePerInstance)
{
var perInst = eval.GetPerInstanceMetrics(dataEval);
perInstance = new RoleMappedData(perInst, dataEval.Schema.GetColumnRoleNames(), opt: true);
}
return(new FoldResult(dict, dataEval.Schema.Schema, perInstance, trainData.Schema));
}
}
IDataTransform AppendToPipeline(IDataView input)
{
IDataView current = input;
if (_shuffleInput)
{
var args1 = new RowShufflingTransformer.Options()
{
ForceShuffle = false,
ForceShuffleSeed = _seedShuffle,
PoolRows = _poolRows,
PoolOnly = false,
};
current = new RowShufflingTransformer(Host, args1, current);
}
// We generate a random number.
var columnName = current.Schema.GetTempColumnName();
var args2 = new GenerateNumberTransform.Options()
{
Columns = new GenerateNumberTransform.Column[] { new GenerateNumberTransform.Column()
{
Name = columnName
} },
Seed = _seed ?? 42
};
IDataTransform currentTr = new GenerateNumberTransform(Host, args2, current);
// We convert this random number into a part.
var cRatios = new float[_ratios.Length];
cRatios[0] = 0;
for (int i = 1; i < _ratios.Length; ++i)
{
cRatios[i] = cRatios[i - 1] + _ratios[i - 1];
}
ValueMapper <float, int> mapper = (in float src, ref int dst) =>
{
for (int i = cRatios.Length - 1; i > 0; --i)
{
if (src >= cRatios[i])
{
dst = i;
return;
}
}
dst = 0;
};
// Get location of columnName
int index = SchemaHelper.GetColumnIndex(currentTr.Schema, columnName);
var ct = currentTr.Schema[index].Type;
var view = LambdaColumnMapper.Create(Host, "Key to part mapper", currentTr,
columnName, _newColumn, ct, NumberDataViewType.Int32, mapper);
// We cache the result to avoid the pipeline to change the random number.
var args3 = new ExtendedCacheTransform.Arguments()
{
inDataFrame = string.IsNullOrEmpty(_cacheFile),
numTheads = _numThreads,
cacheFile = _cacheFile,
reuse = _reuse,
};
currentTr = new ExtendedCacheTransform(Host, args3, view);
// Removing the temporary column.
var objtr = ColumnSelectingTransformer.CreateDrop(Host, currentTr, new string[] { columnName });
var finalTr = objtr as IDataTransform;
if (finalTr == null)
{
throw Contracts.ExceptNotSupp("Desgin change.");
}
var taggedViews = new List <Tuple <string, ITaggedDataView> >();
// filenames
if (_filenames != null || _tags != null)
{
int nbf = _filenames == null ? 0 : _filenames.Length;
if (nbf > 0 && nbf != _ratios.Length)
{
throw Host.Except("Differen number of filenames and ratios.");
}
int nbt = _tags == null ? 0 : _tags.Length;
if (nbt > 0 && nbt != _ratios.Length)
{
throw Host.Except("Differen number of filenames and ratios.");
}
int nb = Math.Max(nbf, nbt);
using (var ch = Host.Start("Split the datasets and stores each part."))
{
for (int i = 0; i < nb; ++i)
{
if (_filenames == null || !_filenames.Any())
{
ch.Info("Create part {0}: {1} (tag: {2})", i + 1, _ratios[i], _tags[i]);
}
else
{
ch.Info("Create part {0}: {1} (file: {2})", i + 1, _ratios[i], _filenames[i]);
}
var ar1 = new RangeFilter.Options()
{
Column = _newColumn, Min = i, Max = i, IncludeMax = true
};
int pardId = i;
var filtView = LambdaFilter.Create <int>(Host, string.Format("Select part {0}", i), currentTr,
_newColumn, NumberDataViewType.Int32,
(in int part) => { return(part.Equals(pardId)); });
