public void TestI_ULabelToR4LabelTransform()
{
var methodName = System.Reflection.MethodBase.GetCurrentMethod().Name;
var dataFilePath = FileHelper.GetTestFile("iris_binary.txt");
var outModelFilePath = FileHelper.GetOutputFile("outModelFilePath.zip", methodName);
var outData = FileHelper.GetOutputFile("outData1.txt", methodName);
var outData2 = FileHelper.GetOutputFile("outData2.txt", methodName);
using (var env = EnvHelper.NewTestEnvironment(conc: 1))
{
var loader = env.CreateLoader("Text{col=LabelText:TX:0 col=Slength:R4:1 col=Swidth:R4:2 col=Plength:R4:3 col=Pwidth:R4:4 header=+}",
new MultiFileSource(dataFilePath));
var concat = env.CreateTransform("Concat{col=Features:Slength,Swidth}", loader);
var labelTx = env.CreateTransform("TermTransform{col=LabelU4:LabelText}", concat);
var labelR4 = env.CreateTransform("U2R4{col=Label:LabelU4}", labelTx);
var roles = env.CreateExamples(labelR4, "Features", "Label");
var trainer = env.CreateTrainer("lr");
using (var ch = env.Start("test"))
{
var pred = trainer.Train(env, ch, roles);
TestTrainerHelper.FinalizeSerializationTest(env, outModelFilePath, pred, roles, outData, outData2,
trainer.Trainer.PredictionKind, true, ratio: 0.8f);
}
}
}
static void TrainPrePostProcessTrainer(string modelName, bool checkError, int threads, bool addpre)
{
var methodName = string.Format("{0}-{1}-T{2}", System.Reflection.MethodBase.GetCurrentMethod().Name, modelName, threads);
var dataFilePath = FileHelper.GetTestFile("mc_iris.txt");
var trainFile = FileHelper.GetOutputFile("iris_train.idv", methodName);
var testFile = FileHelper.GetOutputFile("iris_test.idv", methodName);
var outModelFilePath = FileHelper.GetOutputFile("outModelFilePath.zip", methodName);
var outData = FileHelper.GetOutputFile("outData1.txt", methodName);
var outData2 = FileHelper.GetOutputFile("outData2.txt", methodName);
using (var env = EnvHelper.NewTestEnvironment(conc: threads == 1 ? 1 : 0))
{
var loader = env.CreateLoader("Text{col=Label:R4:0 col=Slength:R4:1 col=Swidth:R4:2 col=Plength:R4:3 col=Pwidth:R4:4 header=+}",
new MultiFileSource(dataFilePath));
var xf = env.CreateTransform("shuffle{force=+}", loader); // We shuffle because Iris is order by label.
xf = env.CreateTransform("concat{col=Features:Slength,Swidth}", xf);
var roles = env.CreateExamples(xf, "Features", "Label");
string pred = addpre ? "PrePost{pre=poly{col=Features} p=___ pret=Take{n=80}}" : "PrePost{p=___ pret=Take{n=80}}";
pred = pred.Replace("___", modelName);
var trainer = env.CreateTrainer(pred);
using (var ch = env.Start("Train"))
{
var predictor = trainer.Train(env, ch, roles);
TestTrainerHelper.FinalizeSerializationTest(env, outModelFilePath, predictor, roles, outData, outData2,
PredictionKind.MultiClassClassification, checkError, ratio: 0.15f);
}
}
}
public static void TrainkNNMultiClassification(int k, NearestNeighborsWeights weight, int threads, float ratio = 0.2f,
string distance = "L2")
{
var methodName = string.Format("{0}-k{1}-W{2}-T{3}-D{4}", System.Reflection.MethodBase.GetCurrentMethod().Name, k, weight, threads, distance);
var dataFilePath = FileHelper.GetTestFile("iris.txt");
var outModelFilePath = FileHelper.GetOutputFile("outModelFilePath.zip", methodName);
var outData = FileHelper.GetOutputFile("outData1.txt", methodName);
var outData2 = FileHelper.GetOutputFile("outData2.txt", methodName);
var env = k == 1 ? EnvHelper.NewTestEnvironment(conc: 1) : EnvHelper.NewTestEnvironment();
using (env)
{
var loader = env.CreateLoader("Text{col=Label:R4:0 col=Slength:R4:1 col=Swidth:R4:2 col=Plength:R4:3 col=Pwidth:R4:4 header=+}",
new MultiFileSource(dataFilePath));
var concat = env.CreateTransform("Concat{col=Features:Slength,Swidth}", loader);
var roles = env.CreateExamples(concat, "Features", "Label");
string modelDef;
modelDef = string.Format("knnmc{{k={0} weighting={1} nt={2} distance={3}}}", k,
weight == NearestNeighborsWeights.distance ? "distance" : "uniform", threads, distance);
var trainer = env.CreateTrainer(modelDef);
using (var ch = env.Start("test"))
{
var pred = trainer.Train(env, ch, roles);
TestTrainerHelper.FinalizeSerializationTest(env, outModelFilePath, pred, roles, outData, outData2,
PredictionKind.MultiClassClassification, true, ratio: ratio);
}
}
}
public static void TrainMultiToRankerPredictorSparse(bool singleColumn, bool checkError)
{
var methodName = string.Format("{0}-{1}-V{2}", System.Reflection.MethodBase.GetCurrentMethod().Name,
"lr", singleColumn ? "C" : "Vec");
var trainFile = FileHelper.GetTestFile("Train-28x28_small.txt");
var testFile = FileHelper.GetTestFile("Test-28x28_small.txt");
var outModelFilePath = FileHelper.GetOutputFile("outModelFilePath.zip", methodName);
var outData = FileHelper.GetOutputFile("outData1.txt", methodName);
var outData2 = FileHelper.GetOutputFile("outData2.txt", methodName);
/*using (*/ var env = EnvHelper.NewTestEnvironment();
{
var loader = env.CreateLoader("Text", new MultiFileSource(trainFile));
var roles = env.CreateExamples(loader, "Features", "Label");
var iova = string.Format("iovark{{p=ftrank sc={0}}}", singleColumn ? "+" : "-");
loader = env.CreateLoader("Text", new MultiFileSource(testFile));
var trainer = env.CreateTrainer(iova);
using (var ch = env.Start("train"))
{
var predictor = trainer.Train(env, ch, roles);
TestTrainerHelper.FinalizeSerializationTest(env, outModelFilePath, predictor, roles, outData, outData2,
PredictionKind.MulticlassClassification, checkError, ratio: 0.1f);
}
}
}
static void TrainMultiToBinaryPredictorSparse(bool singleColumn, bool checkError)
{
var methodName = string.Format("{0}-{1}-V{2}", System.Reflection.MethodBase.GetCurrentMethod().Name,
"lr", singleColumn ? "C" : "Vec");
var trainFile = FileHelper.GetTestFile("Train-28x28_small.txt");
var testFile = FileHelper.GetTestFile("Test-28x28_small.txt");
var outModelFilePath = FileHelper.GetOutputFile("outModelFilePath.zip", methodName);
var outData = FileHelper.GetOutputFile("outData1.txt", methodName);
var outData2 = FileHelper.GetOutputFile("outData2.txt", methodName);
/*using (*/
var env = EnvHelper.NewTestEnvironment(conc: 1);
{
var loader = env.CreateLoader("Text{col=Label:R4:0 col=Features:R4:1-784}", new MultiFileSource(trainFile));
var roles = env.CreateExamples(loader, "Features", "Label");
var df = DataFrameIO.ReadView(roles.Data);
Assert.IsTrue(df.Shape[0] > 0);
var iova = string.Format("iova{{p=lr sc={0} nt=1}}", singleColumn ? "+" : "-");
loader = env.CreateLoader("Text{col=Label:R4:0 col=Features:R4:1-784}", new MultiFileSource(testFile));
var trainer = env.CreateTrainer(iova);
using (var ch = env.Start("train"))
{
var predictor = trainer.Train(env, ch, roles);
TestTrainerHelper.FinalizeSerializationTest(env, outModelFilePath, predictor, roles, outData, outData2,
PredictionKind.MulticlassClassification, checkError, ratio: 0.1f);
}
}
}
static void TrainMultiToRankerPredictorDense(string modelName, int threads, bool checkError,
bool singleColumn, bool shift, bool useUint)
{
var methodName = string.Format("{0}-{1}-V{2}-T{3}-S{4}", System.Reflection.MethodBase.GetCurrentMethod().Name,
modelName, singleColumn ? "C" : "Vec", threads, shift ? "shift" : "std");
var dataFilePath = shift
? FileHelper.GetTestFile("mc_iris_shift.txt")
: FileHelper.GetTestFile("mc_iris.txt");
var trainFile = FileHelper.GetOutputFile("iris_train.idv", methodName);
var testFile = FileHelper.GetOutputFile("iris_test.idv", methodName);
var outModelFilePath = FileHelper.GetOutputFile("outModelFilePath.zip", methodName);
var outData = FileHelper.GetOutputFile("outData1.txt", methodName);
var outData2 = FileHelper.GetOutputFile("outData2.txt", methodName);
/*using (*/
var env = EnvHelper.NewTestEnvironment(conc: threads == 1 ? 1 : 0);
{
string labelType = useUint ? "U4[0-2]" : "R4";
string loadSettings = string.Format("Text{{col=Label:{0}:0 col=Slength:R4:1 col=Swidth:R4:2 col=Plength:R4:3 col=Pwidth:R4:4 header=+}}", labelType);
var loader = env.CreateLoader(loadSettings, new MultiFileSource(dataFilePath));
var concat = env.CreateTransform("Concat{col=Features:Slength,Swidth}", loader);
var roles = env.CreateExamples(concat, "Features", "Label");
string modelDef = threads <= 0 ? modelName : string.Format("{0}{{t={1}}}", modelName, threads);
string additionnal = modelName.Contains("xgbrk") ? " u4=+" : "";
string iova = string.Format("iovark{{p={0} sc={1}{2}}}", modelDef, singleColumn ? "+" : "-", additionnal);
var trainer = env.CreateTrainer(iova);
using (var ch = env.Start("train"))
{
var predictor = trainer.Train(env, ch, roles);
TestTrainerHelper.FinalizeSerializationTest(env, outModelFilePath, predictor, roles, outData, outData2,
PredictionKind.MulticlassClassification, checkError, ratio: 0.1f);
}
}
}
static void OptimizedOVA(float downsampling, string type, string model)
{
var methodName = string.Format("{0}-D{1}-{2}-{3}", System.Reflection.MethodBase.GetCurrentMethod().Name, downsampling, type, model);
string trainFile, testFile;
if (type == "R4")
{
trainFile = FileHelper.GetTestFile("types/iris_train.idv");
testFile = FileHelper.GetTestFile("types/iris_test.idv");
}
else if (type == "U4")
{
trainFile = FileHelper.GetTestFile("types/iris_train_u4.idv");
testFile = FileHelper.GetTestFile("types/iris_test_u4.idv");
}
else if (type == "U43")
{
trainFile = FileHelper.GetTestFile("types/iris_train_u43.idv");
testFile = FileHelper.GetTestFile("types/iris_test_u43.idv");
}
else
{
throw new NotSupportedException();
}
var outModelFilePath = FileHelper.GetOutputFile("outModelFilePath.zip", methodName);
var outData = FileHelper.GetOutputFile("outData1.txt", methodName);
var outData2 = FileHelper.GetOutputFile("outData2.txt", methodName);
StringWriter sout, serr;
/*using (*/
var env = EnvHelper.NewTestEnvironment(out sout, out serr, verbose: false);
{
var loaderSettings = "Binary";
var loader = env.CreateLoader(loaderSettings, new MultiFileSource(trainFile));
var xf = env.CreateTransform("concat{col=Features:Slength,Swidth}", loader);
var roles = env.CreateExamples(xf, "Features", "Label");
var trainer = env.CreateTrainer(string.Format("oova{{p={1} ds={0}}}", downsampling, model));
using (var ch = env.Start("Train"))
{
var pred = trainer.Train(env, ch, roles);
var sbout = sout.GetStringBuilder().ToString();
var sbrr = serr.GetStringBuilder().ToString();
loader = env.CreateLoader(loaderSettings, new MultiFileSource(testFile));
TestTrainerHelper.FinalizeSerializationTest(env, outModelFilePath, pred, roles, outData, outData2,
PredictionKind.MulticlassClassification, true, ratio: 0.8f);
}
}
}
public void TestChainTransformSerializeWithKMeans()
{
var methodName = System.Reflection.MethodBase.GetCurrentMethod().Name;
var dataFilePath = FileHelper.GetTestFile("iris_binary.txt");
var outModelFilePath = FileHelper.GetOutputFile("outModelFilePath.zip", methodName);
var outData = FileHelper.GetOutputFile("outData1.txt", methodName);
var outData2 = FileHelper.GetOutputFile("outData2.txt", methodName);
using (var env = EnvHelper.NewTestEnvironment())
{
var loader = env.CreateLoader("Text{col=Label:R4:0 col=Features:R4:1-4 header=+}",
new MultiFileSource(dataFilePath));
var xf = env.CreateTransform("ChainTrans{xf1=Scaler{col=Features} xf2=Scaler{col=Features}}", loader);
var roles = env.CreateExamples(xf, "Features");
var trainer = env.CreateTrainer("KMeansPlusPlus{k=5}");
using (var ch = env.Start("Train"))
{
var pred = trainer.Train(env, ch, roles);
TestTrainerHelper.FinalizeSerializationTest(env, outModelFilePath, pred, roles, outData, outData2,
PredictionKind.Clustering, false);
}
}
}
static void TrainMultiToBinaryPredictorIris(int th, bool singleColumn, string model, string type)
{
var methodName = string.Format("{0}-T{1}-{2}-{3}-{4}", System.Reflection.MethodBase.GetCurrentMethod().Name, th, singleColumn ? "asvec" : "asR4", model, type);
string trainFile, testFile;
if (type == "R4")
{
trainFile = FileHelper.GetTestFile("types/iris_train.idv");
testFile = FileHelper.GetTestFile("types/iris_test.idv");
}
else if (type == "U4")
{
trainFile = FileHelper.GetTestFile("types/iris_train_u4.idv");
testFile = FileHelper.GetTestFile("types/iris_test_u4.idv");
}
else if (type == "U43")
{
trainFile = FileHelper.GetTestFile("types/iris_train_u43.idv");
testFile = FileHelper.GetTestFile("types/iris_test_u43.idv");
}
else
{
throw new NotSupportedException();
}
var outModelFilePath = FileHelper.GetOutputFile("outModelFilePath.zip", methodName);
var outData = FileHelper.GetOutputFile("outData1.txt", methodName);
var outData2 = FileHelper.GetOutputFile("outData2.txt", methodName);
/*using (*/
var env = EnvHelper.NewTestEnvironment(conc: th == 1 ? 1 : 0);
{
var loaderSettings = "Binary";
var loader = env.CreateLoader(loaderSettings, new MultiFileSource(trainFile));
var xf = env.CreateTransform("concat{col=Features:Slength,Swidth}", loader);
var roles = env.CreateExamples(xf, "Features", "Label");
ITrainerExtended trainer;
if (model.ToLower() == "ova" || model.ToLower() == "oova")
{
if (th > 0)
{
trainer = env.CreateTrainer(string.Format("oova{{ p=ft{{t={0}}} }}", th, singleColumn ? "+" : "-"));
}
else
{
trainer = env.CreateTrainer(string.Format("oova{{p=ft{{t=1}} }}", singleColumn ? "+" : "-"));
}
}
else
{
if (th > 0)
{
trainer = env.CreateTrainer(string.Format("iova{{ p=ft{{t={0}}} sc={1} }}", th, singleColumn ? "+" : "-"));
}
else
{
trainer = env.CreateTrainer(string.Format("iova{{p=ft{{t=1}} sc={0} }}", singleColumn ? "+" : "-"));
}
}
using (var ch = env.Start("Train"))
{
var pred = trainer.Train(env, ch, roles);
loader = env.CreateLoader(loaderSettings, new MultiFileSource(testFile));
TestTrainerHelper.FinalizeSerializationTest(env, outModelFilePath, pred, roles, outData, outData2,
PredictionKind.MulticlassClassification, true,
ratio: type.StartsWith("U4") && model.ToLower() == "iova" ? 1f : 0.1f);
}
}
}
