public void TestWizard()
{
FileInfo rawFile = TEMP_DIR.CreateFile("iris_raw.csv");
FileUtil.CopyResource("Encog.Resources.iris.csv", rawFile);
FileInfo analystFile = TEMP_DIR.CreateFile("iris.ega");
EncogAnalyst encog = new EncogAnalyst();
AnalystWizard wiz = new AnalystWizard(encog);
wiz.Goal = AnalystGoal.Classification;
wiz.Wizard(rawFile, true, AnalystFileFormat.DecpntComma);
encog.ExecuteTask("task-full");
encog.Save(analystFile);
encog.Load(analystFile);
AnalystReport report = new AnalystReport(encog);
report.ProduceReport(TEMP_DIR.CreateFile("report.html"));
Assert.AreEqual(5, encog.Script.Normalize.NormalizedFields.Count);
Assert.AreEqual(4.3, encog.Script.Fields[0].Min, 0.001);
Assert.AreEqual(7.9, encog.Script.Fields[0].Max, 0.001);
Assert.AreEqual(5.8483221477, encog.Script.Fields[0].Mean, 0.001);
Assert.AreEqual(encog.Script.Fields[0].Class, false);
Assert.AreEqual(encog.Script.Fields[0].Real, true);
Assert.AreEqual(encog.Script.Fields[0].Integer, false);
Assert.AreEqual(encog.Script.Fields[0].Complete, true);
Assert.AreEqual(-3.38833, encog.Script.Normalize.NormalizedFields[0].Normalize(0.001), 0.001);
}
public void Execute(IExampleInterface app)
{
if (app.Args.Length != 2)
{
Console.WriteLine(@"Note: This example assumes that headers are present in the CSV files.");
Console.WriteLine(@"NormalizeFile [input file] [target file]");
}
else
{
var sourceFile = new FileInfo(app.Args[0]);
var targetFile = new FileInfo(app.Args[1]);
var analyst = new EncogAnalyst();
var wizard = new AnalystWizard(analyst);
wizard.Wizard(sourceFile, true, AnalystFileFormat.DecpntComma);
DumpFieldInfo(analyst);
var norm = new AnalystNormalizeCSV();
norm.Analyze(sourceFile, true, CSVFormat.English, analyst);
norm.ProduceOutputHeaders = true;
norm.Normalize(targetFile);
EncogFramework.Instance.Shutdown();
}
}
public static IMLDataSet LoadAndNormalizeData(FileInfo fileInfo, AnalystGoal problemType, NormalizationAction normalizationType, bool randomize = true)
{
var analyst = new EncogAnalyst();
var wizard = new AnalystWizard(analyst);
wizard.Goal = problemType;
wizard.Wizard(fileInfo, true, AnalystFileFormat.DecpntComma);
var fields = analyst.Script.Normalize.NormalizedFields;
if (problemType == AnalystGoal.Classification)
fields[fields.Count - 1].Action = normalizationType;
var norm = new AnalystNormalizeCSV();
norm.Analyze(fileInfo, true, CSVFormat.DecimalPoint, analyst);
var normalizedDataFileInfo = new FileInfo("temp/temp.csv");
norm.Normalize(normalizedDataFileInfo);
var inputNeurons = fields.Count - 1;
int outputNeurons;
if (problemType == AnalystGoal.Classification)
outputNeurons = fields.Last().Classes.Count - (normalizationType == NormalizationAction.Equilateral ? 1 : 0);
else
outputNeurons = fields.Count - inputNeurons;
var result = CSVHelper.LoadCSVToDataSet(normalizedDataFileInfo, inputNeurons, outputNeurons, randomize);
normalizedDataFileInfo.Delete();
return result;
}
private static void Normalise()
{
var analyst = new EncogAnalyst();
var wizard = new AnalystWizard(analyst);
wizard.Wizard(sourceFile, true, AnalystFileFormat.DecpntComma);
//for numerical vals:
analyst.Script.Normalize.NormalizedFields[0].Action = Encog.Util.Arrayutil.NormalizationAction.Normalize;
//for enumerations:
analyst.Script.Normalize.NormalizedFields[0].Action = Encog.Util.Arrayutil.NormalizationAction.Equilateral;
}
public static void Normalise(FileOps fileOps)
{
var analyst = new EncogAnalyst();
var wizard = new AnalystWizard(analyst);
wizard.Wizard(fileOps.BaseFile,true,AnalystFileFormat.DecpntComma);
var norm = new AnalystNormalizeCSV{ProduceOutputHeaders = true};
norm.Analyze(fileOps.TrainingFile, true, CSVFormat.English, analyst);
norm.Normalize(fileOps.NormalisedTrainingFile);
norm.Analyze(fileOps.EvaluationFile, true, CSVFormat.English, analyst);
norm.Normalize(fileOps.NormalisedEvaluationFile);
analyst.Save(fileOps.AnalystFile);
}
private void dataNormalization(string f1, string f2)
{
var sourceFile = new FileInfo(f1);
var targetFile = new FileInfo(f2);
var analyst = new EncogAnalyst();
var wizard = new AnalystWizard(analyst);
wizard.Wizard(sourceFile, true, AnalystFileFormat.DecpntComma);
var norm = new AnalystNormalizeCSV();
norm.Analyze(sourceFile, true, CSVFormat.English, analyst);
norm.ProduceOutputHeaders = true;
norm.Normalize(targetFile);
analyst.Save(new FileInfo("stats.ega"));
analyst.Load(new FileInfo("stats.ega"));
}
public void ForestExample(FileInfo dir)
{
Console.WriteLine("Starting forest cover dataset example.");
Uri url = new Uri(FOREST_SOURCE);
FileInfo analystFile = FileUtil.CombinePath(dir, "forest.ega");
FileInfo rawFile = FileUtil.CombinePath(dir, "forest_raw.csv");
EncogAnalyst encog = new EncogAnalyst();
encog.AddAnalystListener(new ConsoleAnalystListener());
AnalystWizard wiz = new AnalystWizard(encog);
wiz.TaskBalance = true;
wiz.Wizard(url, analystFile, rawFile, false, AnalystFileFormat.DecpntComma);
encog.ExecuteTask("task-full");
encog.Save(analystFile);
AnalystReport report = new AnalystReport(encog);
report.ProduceReport(FileUtil.CombinePath(dir, "report.html"));
}
public void IrisExample(FileInfo dir)
{
Console.WriteLine("Starting Iris dataset example.");
Uri url = new Uri(IRIS_SOURCE);
FileInfo analystFile = FileUtil.CombinePath(dir, "iris.ega");
FileInfo rawFile = FileUtil.CombinePath(dir, "iris_raw.csv");
EncogAnalyst encog = new EncogAnalyst();
encog.AddAnalystListener(new ConsoleAnalystListener());
AnalystWizard wiz = new AnalystWizard(encog);
//wiz.TaskRandomize = false;
wiz.Wizard(url, analystFile, rawFile, false, AnalystFileFormat.DecpntComma);
encog.Save(analystFile);
encog.ExecuteTask("task-full");
AnalystReport report = new AnalystReport(encog);
report.ProduceReport(FileUtil.CombinePath(dir, "report.html"));
}
/// <summary>
/// Program entry point.
/// </summary>
/// <param name="app">Holds arguments and other info.</param>
public void Execute(IExampleInterface app)
{
Console.WriteLine("Running wizard...");
var analyst = new EncogAnalyst();
var wizard = new AnalystWizard(analyst);
wizard.TargetFieldName = "field:1";
wizard.Wizard(sourceCSV,
false, AnalystFileFormat.DecpntComma);
// customer id
analyst.Script.Normalize.NormalizedFields[0].Action = Encog.Util.Arrayutil.NormalizationAction.PassThrough;
var norm = new AnalystNormalizeCSV();
norm.Report = new ConsoleStatusReportable();
Console.WriteLine("Analyze for normalize...");
norm.Analyze(sourceCSV, false, CSVFormat.English, analyst);
norm.ProduceOutputHeaders = true;
Console.WriteLine("Normalize...");
norm.Normalize(targetCSV);
analyst.Save(scriptEGA);
}
public void Wizard(AnalystGoal goal, WizardMethodType methodType, bool headers)
{
EncogAnalyst.MaxIteration = MaxIterations;
var wiz = new AnalystWizard(EncogAnalyst) {Goal = goal, MethodType = methodType, EvidenceSegements = 3};
wiz.Wizard(_rawFile, headers, FileFormat);
EncogAnalyst.Save(_analystFile);
EncogAnalyst.Load(_analystFile);
}
/// <summary>
/// Metodo responsavel por normalizar as informacoes para adequar a execucao da rede neural
/// </summary>
private static void Normalization()
{
var analyst = new EncogAnalyst();
var wizard = new AnalystWizard(analyst);
wizard.Wizard(Config.ClassificationFile, true, AnalystFileFormat.DecpntComma);
var norm = new AnalystNormalizeCSV();
norm.Analyze(Config.TrainingClassificationFile, true, CSVFormat.English, analyst);
norm.ProduceOutputHeaders = true;
norm.Normalize(Config.NormalizedTrainingClassificationFile);
norm.Analyze(Config.EvaluateClassificationFile, true, CSVFormat.English, analyst);
norm.Normalize(Config.NormalizedEvaluateClassificationFile);
analyst.Save(Config.AnalystClassificationFile);
}
public void Analyze(TextBox txtOut)
{
// Use analyst
m_analyst = new EncogAnalyst();
var source = new FileInfo(m_fileName);
var wizard = new AnalystWizard(m_analyst);
wizard.Wizard(source, true, AnalystFileFormat.DecpntComma);
// Set the minimum value for normalization
foreach (AnalystField field in m_analyst.Script.Normalize.NormalizedFields)
field.NormalizedLow = 0.0;
txtOut.AppendText("Fields found in file:" + Environment.NewLine);
foreach (AnalystField field in m_analyst.Script.Normalize.NormalizedFields)
txtOut.AppendText(String.Format("{0}, action = {1}, min = {2}, max = {3}{4}",
field.Name, field.Action, field.ActualLow, field.ActualHigh, Environment.NewLine));
}
private static void Normalize(string sourceFile, string targetFile)
{
var sourceFileInfo = new FileInfo(sourceFile);
var targetFileInfo = new FileInfo(targetFile);
var analyst = new EncogAnalyst();
var wizard = new AnalystWizard(analyst);
wizard.Goal = AnalystGoal.Unknown; // Default Classification으로하면 Wizard 동작 안함
wizard.Wizard(sourceFileInfo, true, AnalystFileFormat.DecpntComma);
var norm = new AnalystNormalizeCSV();
norm.Analyze(sourceFileInfo, true, CSVFormat.English, analyst);
norm.ProduceOutputHeaders = true;
norm.Normalize(targetFileInfo);
}
/// <summary>
/// Metodo responsavel por normalizar as informacoes para adequar a execucao da rede neural
/// </summary>
private static void Normalization()
{
var analyst = new EncogAnalyst();
//Wizard
var wizard = new AnalystWizard(analyst);
wizard.Wizard(Config.RegressionFile, true, AnalystFileFormat.DecpntComma);
//Cilindros
analyst.Script.Normalize.NormalizedFields[0].Action = Encog.Util.Arrayutil.NormalizationAction.Equilateral;
//displacement
analyst.Script.Normalize.NormalizedFields[1].Action = Encog.Util.Arrayutil.NormalizationAction.Normalize;
//HorsePower
analyst.Script.Normalize.NormalizedFields[2].Action = Encog.Util.Arrayutil.NormalizationAction.Normalize;
//Peso
analyst.Script.Normalize.NormalizedFields[3].Action = Encog.Util.Arrayutil.NormalizationAction.Normalize;
//Aceleração
analyst.Script.Normalize.NormalizedFields[4].Action = Encog.Util.Arrayutil.NormalizationAction.Normalize;
//Ano
analyst.Script.Normalize.NormalizedFields[5].Action = Encog.Util.Arrayutil.NormalizationAction.Equilateral;
//Origem
analyst.Script.Normalize.NormalizedFields[6].Action = Encog.Util.Arrayutil.NormalizationAction.Equilateral;
//Nome
analyst.Script.Normalize.NormalizedFields[7].Action = Encog.Util.Arrayutil.NormalizationAction.Ignore;
//MPG
analyst.Script.Normalize.NormalizedFields[8].Action = Encog.Util.Arrayutil.NormalizationAction.Normalize;
var norm = new AnalystNormalizeCSV();
norm.ProduceOutputHeaders = true;
norm.Analyze(Config.TrainingRegressionFile, true, CSVFormat.English, analyst);
norm.Normalize(Config.NormalizedTrainingRegressionFile);
//Norm of evaluation
norm.Analyze(Config.EvaluateRegressionFile, true, CSVFormat.English, analyst);
norm.Normalize(Config.NormalizedEvaluateRegressionFile);
//save the analyst file
analyst.Save(Config.AnalystRegressionFile);
}
private void WizardCommand()
{
String targetCSVFile = _cmd.Args[0];
String egaFile = FileUtil.ForceExtension(targetCSVFile, "ega");
var analyst = new EncogAnalyst();
var wizard = new AnalystWizard(analyst);
bool headers = _cmd.PromptBoolean("headers", true);
AnalystFileFormat format =
ConvertStringConst.String2AnalystFileFormat(_cmd.PromptString("format", "decpnt|comma"));
String goal = _cmd.PromptString("goal", "c").ToLower().Trim();
if (goal.Equals("c"))
{
wizard.Goal = AnalystGoal.Classification;
}
else if (goal.Equals("r"))
{
wizard.Goal = AnalystGoal.Regression;
}
else
{
Console.WriteLine(@"Invalid goal: " + goal);
return;
}
wizard.TargetField = _cmd.PromptString("targetField", "");
String m = _cmd.PromptString("method", "ff").ToLower();
if (m.Equals("ff"))
{
wizard.MethodType = WizardMethodType.FeedForward;
}
else if (m.Equals("neat"))
{
wizard.MethodType = WizardMethodType.NEAT;
}
else if (m.Equals("pnn"))
{
wizard.MethodType = WizardMethodType.PNN;
}
else if (m.Equals("rbf"))
{
wizard.MethodType = WizardMethodType.RBF;
}
else if (m.Equals("som"))
{
wizard.MethodType = WizardMethodType.SOM;
}
else if (m.Equals("svm"))
{
wizard.MethodType = WizardMethodType.SVM;
}
else
{
Console.WriteLine(@"Invalid method: " + m);
return;
}
String r = _cmd.PromptString("range", "-1t1").Trim().ToLower();
if (r.Equals("-1t1"))
{
wizard.Range = NormalizeRange.NegOne2One;
}
else if (r.Equals("0t1"))
{
wizard.Range = NormalizeRange.Zero2One;
}
wizard.Missing = TranslateMissing(_cmd.PromptString("missing", "DiscardMissing"));
wizard.LagWindowSize = _cmd.PromptInteger("lagWindow", 0);
wizard.LeadWindowSize = _cmd.PromptInteger("leadWindow", 0);
wizard.IncludeTargetField = _cmd.PromptBoolean("includeTarget", false);
wizard.TaskNormalize = _cmd.PromptBoolean("normalize", true);
wizard.TaskRandomize = _cmd.PromptBoolean("randomize", true);
wizard.TaskSegregate = _cmd.PromptBoolean("segregate", true);
wizard.TaskBalance = _cmd.PromptBoolean("balance", false);
wizard.TaskCluster = _cmd.PromptBoolean("cluster", false);
_sw.Start();
Console.WriteLine(@"Analyzing data");
wizard.Wizard(new FileInfo(targetCSVFile), headers, format);
Console.WriteLine(@"Saving analyst file");
analyst.Save(egaFile);
}