/// <summary>
/// Initialize NBN with data from specified file
/// </summary>
/// <param name="filename">String - filename</param>
/// <returns>bool - success flag</returns>
private bool loadInputData(String filename)
{
try
{
if (!File.Exists(filename))
{
throw new NeuralNetworkError(String.Format(Properties.Settings.Default.NBN1, filename));
}
try
{
var data = MatrixMB.Load(filename);
if (MatLabCompareDataFolder.Length > 0)
{
//data from matlab are normalized
string name = Path.GetFileNameWithoutExtension(Directory.GetFiles(MatLabCompareDataFolder, "*.dat")[0]);
inputLearn = MatrixMB.Load(string.Format("{0}\\uczenie_wejscie_{1}.txt", MatLabCompareDataFolder, name)).ToInput();
inputTest = MatrixMB.Load(string.Format("{0}\\testowanie_wejscie_{1}.txt", MatLabCompareDataFolder, name)).ToInput();
outputLearn = MatrixMB.Load(string.Format("{0}\\uczenie_wyjscie_{1}.txt", MatLabCompareDataFolder, name)).ToOutput();
outputTest = MatrixMB.Load(string.Format("{0}\\testowanie_wyjscie_{1}.txt", MatLabCompareDataFolder, name)).ToOutput();
}
else
{
input = data.CopyColumnsWithoutLast().ToInput();
input.Normalize();
output = data.LastColumn.ToOutput();
output.Normalize();
if (IsClassification)
{
int Tnp = data.Rows;
inputLearn = input.CopyRows(Tnp - 1).ToInput();
inputTest = input.CopyRows(Tnp - 1).ToInput();
outputLearn = output.CopyRows(Tnp - 1).ToOutput();
outputTest = output.CopyRows(Tnp - 1).ToOutput();
}
else
{
int[] ind = data.Rows.RandomPermutation();
int Tnp = Math.Round(data.Rows * 0.7).ToInt();
inputTest = input.CopyRows(Tnp, input.Rows - 1).ToInput();
inputLearn = input.CopyRows(Tnp - 1).ToInput();
outputTest = output.CopyRows(Tnp, input.Rows - 1).ToOutput();
outputLearn = output.CopyRows(Tnp - 1).ToOutput();
}
}
try
{
if (IsResearchMode)//export for matlab - out of use
{
string name = Path.GetFileNameWithoutExtension(filename);
DateTime d = DateTime.Now;
string dir = String.Format("{0}\\Executed research\\{1}\\{2}_{3}_{4}_{5}_{6}_{7}",
Path.GetDirectoryName(filename), name,
d.Year, d.Month, d.Day, d.Hour, d.Minute, d.Second);
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
extractionFolder = dir;
_reasearch_folder = dir;
string learn_input = String.Format("{0}\\{1}_learn_input.dat", dir, name);
string test_input = String.Format("{0}\\{1}_test_input.dat", dir, name);
string learn_output = String.Format("{0}\\{1}_learn_output.dat", dir, name);
string test_output = String.Format("{0}\\{1}_test_output.dat", dir, name);
string initialWweights = String.Format("{0}\\{1}_initial_weights.dat", dir, name);
inputTest.Store(test_input);
inputLearn.Store(learn_input);
outputTest.Store(test_output);
outputLearn.Store(learn_output);
}
}
catch (Exception)
{ }
}
catch (Exception iex)
{
throw new NeuralNetworkError(String.Format(Properties.Settings.Default.NBN2, filename), iex);
}
return(true);
}
catch (Exception ex)
{
updateErrorNBN(ex);
return(false);
}
}
/// <summary>
/// Initialize NBN with data from specified file
/// </summary>
/// <param name="filename">String - filename</param>
/// <returns>bool - success flag</returns>
private bool loadInputData(String filename)
{
try
{
if (!File.Exists(filename))
{
throw new NeuralNetworkError(String.Format(Properties.Settings.Default.NBN1, filename));
}
try
{
var data = MatrixMB.Load(filename);
if (MatLabCompareDataFolder.Length > 0)
{
//data from matlab are normalized
string name = Path.GetFileNameWithoutExtension(Directory.GetFiles(MatLabCompareDataFolder, "*.dat")[0]);
inputLearn = MatrixMB.Load(string.Format("{0}\\uczenie_wejscie_{1}.txt", MatLabCompareDataFolder, name)).ToInput();
inputTest = MatrixMB.Load(string.Format("{0}\\testowanie_wejscie_{1}.txt", MatLabCompareDataFolder, name)).ToInput();
outputLearn = MatrixMB.Load(string.Format("{0}\\uczenie_wyjscie_{1}.txt", MatLabCompareDataFolder, name)).ToOutput();
outputTest = MatrixMB.Load(string.Format("{0}\\testowanie_wyjscie_{1}.txt", MatLabCompareDataFolder, name)).ToOutput();
}
else
{
input = data.CopyColumnsWithoutLast().ToInput();
input.Normalize();
output = data.LastColumn.ToOutput();
output.Normalize();
if (IsClassification)
{
int Tnp = data.Rows;
inputLearn = input.CopyRows(Tnp - 1).ToInput();
inputTest = input.CopyRows(Tnp - 1).ToInput();
outputLearn = output.CopyRows(Tnp - 1).ToOutput();
outputTest = output.CopyRows(Tnp - 1).ToOutput();
}
else
{
int[] ind = data.Rows.RandomPermutation();
int Tnp = Math.Round(data.Rows * 0.7).ToInt();
inputTest = input.CopyRows(Tnp, input.Rows - 1).ToInput();
inputLearn = input.CopyRows(Tnp - 1).ToInput();
outputTest = output.CopyRows(Tnp, input.Rows - 1).ToOutput();
outputLearn = output.CopyRows(Tnp - 1).ToOutput();
}
}
try
{
if (IsResearchMode)//export for matlab - out of use
{
string name = Path.GetFileNameWithoutExtension(filename);
DateTime d = DateTime.Now;
string dir = String.Format("{0}\\Executed research\\{1}\\{2}_{3}_{4}_{5}_{6}_{7}",
Path.GetDirectoryName(filename), name,
d.Year, d.Month, d.Day, d.Hour, d.Minute, d.Second);
if (!Directory.Exists(dir)) Directory.CreateDirectory(dir);
extractionFolder = dir;
_reasearch_folder = dir;
string learn_input = String.Format("{0}\\{1}_learn_input.dat", dir, name);
string test_input = String.Format("{0}\\{1}_test_input.dat", dir, name);
string learn_output = String.Format("{0}\\{1}_learn_output.dat", dir, name);
string test_output = String.Format("{0}\\{1}_test_output.dat", dir, name);
string initialWweights = String.Format("{0}\\{1}_initial_weights.dat", dir, name);
inputTest.Store(test_input);
inputLearn.Store(learn_input);
outputTest.Store(test_output);
outputLearn.Store(learn_output);
}
}
catch (Exception)
{ }
}
catch (Exception iex)
{
throw new NeuralNetworkError(String.Format(Properties.Settings.Default.NBN2, filename), iex);
}
return true;
}
catch (Exception ex)
{
updateErrorNBN(ex);
return false;
}
}
