public static void Run(string trainingSetPath, string testingSetPath, int numberOfModels, int?randomSeed)
{
Trace.TraceInformation("Starting bagging");
Trace.TraceInformation("TrainingSetPath: {0}", trainingSetPath);
Trace.TraceInformation("TestingSetPath: {0}", testingSetPath);
Trace.TraceInformation("Models: {0}", numberOfModels);
Instances trainingInstances = Diabetes.LoadData(trainingSetPath, Mode.Train);
Bagging bagging = new Bagging();
bagging.Train(trainingInstances, numberOfModels, randomSeed);
Instances testingInstances = Diabetes.LoadData(testingSetPath, Mode.Test);
bagging.TestNonBagging(testingInstances);
bagging.Test(testingInstances);
}
public static void GenerateNaiveBayesInputFiles(string trainingInputFile, string testingInputFile, string trainingOutputFile, string testingOutputFile)
{
// We don't want to change the naive bayes implementation for spam detection. So, let target attribute 0 = "ham" and 1 = "spam"
// Training Data
Instances trainingInstances = Diabetes.LoadData(trainingInputFile, Mode.Train, Category.Buckets);
int targetAttributeIndex = Diabetes.NumAttributes - 1;
StringBuilder trainingOutput = new StringBuilder();
for (int i = 0; i < trainingInstances.numInstances(); i++)
{
trainingOutput.AppendFormat("{0} {1}", i, trainingInstances.instance(i).value(targetAttributeIndex) == 0 ? "ham" : "spam");
for (int j = 0; j < trainingInstances.numAttributes() - 1; j++)
{
trainingOutput.AppendFormat(" {0}_{1} {2}", j, trainingInstances.instance(i).value(j), 1);
//trainingOutput.AppendFormat(" {0} {1}", j, trainingInstances.instance(i).value(j));
}
trainingOutput.AppendLine();
}
File.WriteAllText(trainingOutputFile, trainingOutput.ToString());
// Testing Data
Instances testingInstances = Diabetes.LoadData(testingInputFile, Mode.Test, Category.Buckets);
StringBuilder testingOutput = new StringBuilder();
for (int i = 0; i < testingInstances.numInstances(); i++)
{
testingOutput.AppendFormat("{0} {1}", i, testingInstances.instance(i).value(targetAttributeIndex) == 0 ? "ham" : "spam");
for (int j = 0; j < testingInstances.numAttributes() - 1; j++)
{
testingOutput.AppendFormat(" {0}_{1} {2}", j, testingInstances.instance(i).value(j), 1);
//testingOutput.AppendFormat(" {0} {1}", j, testingInstances.instance(i).value(j));
}
testingOutput.AppendLine();
}
File.WriteAllText(testingOutputFile, testingOutput.ToString());
}