//////////////////////////////////////////////////////
// Function:
// LoadTrainingData
//
// Purpose:
// Loads the training data used to classify queries.
//////////////////////////////////////////////////////
public bool LoadTrainingData(
string szInFileName)
{
using (var pInputFile = new StreamReader(szInFileName))
{
int iNumClasses = 0;
int i = 0;
int j = 0;
string szInputString;
// Read in the classes
string input = pInputFile.ReadLine();
input = pInputFile.ReadLine();
iNumClasses = int.Parse(input);
//fscanf(pInputFile, "%d", &iNumClasses);
m_rwExamples.Clear();
m_rwAttributes.Clear();
m_rwClasses.Clear();
// Print out the names of the classes as we read them in
//cout << "Read " << iNumClasses << " classes";
// Read in the classifications
// that we can put the data into
for (i = 0; i < iNumClasses; ++i)
{
szInputString = pInputFile.ReadLine();
//fscanf(pInputFile, "%s", szInputString);
m_rwClasses.Add(szInputString.Trim());
//cout << " " << m_rwClasses[i];
}
//cout << endl;
// Read in the attributes
pInputFile.ReadLine();
//fscanf(pInputFile, "%s", szInputString);
int iNumAttributes = int.Parse(pInputFile.ReadLine());
//fscanf(pInputFile, "%d", &iNumAttributes);
//cout << "Read " << iNumAttributes << " attributes\n";
// Read in each attribute and it's values
for (i = 0; i < iNumAttributes; ++i)
{
// Check to make sure we haven't hit
// a bad file and EOF
if (pInputFile.EndOfStream)
{
return(false);
}
szInputString = pInputFile.ReadLine().Trim();
//fscanf(pInputFile, "%s", szInputString);
m_rwAttributes.Add(new CAttribute());
m_rwAttributes[i].SetName(szInputString);
int iNumValues = int.Parse(pInputFile.ReadLine());
//fscanf(pInputFile, "%d", &iNumValues);
//cout << "Attribute " << m_rwAttributes[i].GetName() << " has " << iNumValues << " values:";
for (j = 0; j < iNumValues; ++j)
{
szInputString = pInputFile.ReadLine().Trim();
//fscanf(pInputFile, "%s", szInputString);
m_rwAttributes[i].AddValue(szInputString);
//cout << " " << m_rwAttributes[i].GetValue(j);
}
//cout << endl;
}
// We should get the keyword "examples"
szInputString = pInputFile.ReadLine().Trim();
//fscanf(pInputFile, "%s", szInputString);
//cout << "EXAMPLES\n";
// Read in the examples until we hit EOF
// If we are running a debug build, print out
// the example data as we read it in
while (!pInputFile.EndOfStream)
{
var newExample = new CExample();
szInputString = pInputFile.ReadLine().Trim();
//int iScanResult(fscanf(pInputFile, "%s", szInputString ) );
if (string.IsNullOrEmpty(szInputString))
{
break;
}
//cout << szInputString;
int iClassIdentifier = 0;
if (!GetClassIdentifier(szInputString, ref iClassIdentifier))
{
Debug.Fail("Unable to find class identifier!");
break;
}
newExample.SetClassIdentifier(iClassIdentifier);
for (i = 0; i < m_rwAttributes.Count; ++i)
{
szInputString = pInputFile.ReadLine().Trim();
//fscanf(pInputFile, "%s", szInputString);
//cout << " " << szInputString;
int iValueIdentifier = 0;
if (m_rwAttributes[i].GetValueIdentifier(szInputString, ref iValueIdentifier))
{
newExample.SetValueIdentifier(i, iValueIdentifier);
}
}
// cout << endl;
// The the newly read example into the example pool.
AddExample(newExample);
// cout << "Read new example: " << newExample << endl;
}
return(true);
}
}
//////////////////////////////////////////////////////
// Mutator:
// AddExample
//
// Purpose:
// Adds an example to the pool of training data.
//////////////////////////////////////////////////////
public void AddExample(
CExample inNewExample)
{
m_rwExamples.Add(inNewExample);
}