private void TrainLOOCVNetworks()
{
m_AllFeaturesForLOOCV = new Dictionary <string, FeatureSet[]>();
try
{
// Read in each source file and create FeatureSets
for (int i = 0; i < m_Source.Count; i++)
{
toolStripStatusLabel1.Text = "Reading Feature File " + i.ToString() + " of " + m_Source.Count.ToString();
FeatureSet[] sets = FeatureSet.GetFeatureSetsFromFile(m_Source[i]);
string str = System.IO.Path.GetFileNameWithoutExtension(m_Source[i]);
if (!m_AllFeaturesForLOOCV.ContainsKey(str))
{
m_AllFeaturesForLOOCV.Add(str, sets);
}
if (!m_QueueForLOOCV.Contains(str))
{
m_QueueForLOOCV.Enqueue(str);
}
}
}
catch (Exception e)
{
MessageBox.Show(e.Message);
return;
}
try
{
m_Worker.RunWorkerAsync(m_QueueForLOOCV.Dequeue());
}
catch (Exception e)
{
MessageBox.Show(e.Message);
}
}
private void buttonTest_Click(object sender, EventArgs e)
{
// Get the trained networks
bool successful;
List <string> networkFiles = General.SelectOpenFiles(out successful, "Select the trained networks to use.",
"Trained Networks (*.nn)|*.nn");
if (!successful)
{
MessageBox.Show("Unable to get network filenames");
return;
}
// Load all the networks from file
Dictionary <string, NeuralNetwork> networks = new Dictionary <string, NeuralNetwork>(networkFiles.Count);
foreach (string file in networkFiles)
{
string nameShort = Path.GetFileNameWithoutExtension(file);
int start = nameShort.IndexOf("Network") + "Network".Length;
string username = nameShort.Substring(start);
NeuralNetwork net = NeuralNetwork.LoadNetworkFromFile(file);
if (!networks.ContainsKey(username))
{
networks.Add(username, net);
}
else
{
MessageBox.Show("Unable to load network " + nameShort);
}
}
// Get the filename to save results to
bool saveFileSuccess;
string saveFilename = General.SelectSaveFile(out saveFileSuccess, "File to save results to", "Text (*.txt)|*.txt");
if (!saveFileSuccess)
{
MessageBox.Show("Unable to select file to save the results to");
return;
}
// Run each file through the networks
Dictionary <string, NeuralNetworkResults> allResults = new Dictionary <string, NeuralNetworkResults>(networks.Count);
foreach (string source in m_Source)
{
string nameShort = Path.GetFileNameWithoutExtension(source);
int start = nameShort.IndexOf("values") + "values".Length + 1;
string user = nameShort.Substring(start);
if (networks.ContainsKey(user))
{
NeuralNetwork net = networks[user];
FeatureSet[] set = FeatureSet.GetFeatureSetsFromFile(source);
bool success;
NeuralNetworkResults results = net.Evaluate(set, "", out success);
if (success)
{
if (!allResults.ContainsKey(user))
{
allResults.Add(user, results);
}
else
{
MessageBox.Show("Already added results for user " + user);
}
}
else
{
MessageBox.Show("Unable to evaluate with user " + user);
}
}
else
{
MessageBox.Show("User: "******" does not have a definition in 'networks'");
}
}
StreamWriter writer = new StreamWriter(saveFilename);
foreach (KeyValuePair <string, NeuralNetworkResults> entry in allResults)
{
int[][] matrix = entry.Value.ConfusionMatrix;
writer.WriteLine(entry.Key + "," + entry.Value.Accuracy.ToString("#0.000") + ","
+ matrix[0][0] + "," + matrix[0][1] + "," + matrix[1][0] + "," + matrix[1][1]);
/*writer.WriteLine(entry.Key + "," + entry.Value.Accuracy.ToString("#0.000") + ",Confusion Matrix,,Classified As");
* writer.WriteLine(",,,,Different Shape, Same Shape");
*
* writer.WriteLine(",,Should Be,Different Shape,{0},{1}", matrix[0][0], matrix[0][1]);
* writer.WriteLine(",,,Same Shape,{0},{1}", matrix[1][0], matrix[1][1]);
*
* writer.WriteLine();*/
}
writer.Close();
}